University of Calgary
PRISM: University of Calgary's Digital Repository
Graduate Studies The Vault: Electronic Theses and Dissertations
2014-09-30
Integration of Geomechanical Parameters and
Numerical Simulation for an Offshore Reservoir in the
Gulf of Mexico
Manzano Angeles, David
Manzano Angeles, D. (2014). Integration of Geomechanical Parameters and Numerical Simulation
for an Offshore Reservoir in the Gulf of Mexico (Unpublished master's thesis). University of
Calgary, Calgary, AB. doi:10.11575/PRISM/26583
http://hdl.handle.net/11023/1862
master thesis
University of Calgary graduate students retain copyright ownership and moral rights for their
thesis. You may use this material in any way that is permitted by the Copyright Act or through
licensing that has been assigned to the document. For uses that are not allowable under
copyright legislation or licensing, you are required to seek permission.
Downloaded from PRISM: https://prism.ucalgary.ca
UNIVERSITY OF CALGARY
Integration of Geomechanical Parameters and Numerical Simulation for an Offshore Reservoir
in the Gulf of Mexico
by
David Manzano Angeles
A THESIS
SUBMITTED TO THE FACULTY OF GRADUATE STUDIES
IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE
DEGREE OF MASTER OF ENGINEERING
GRADUATE PROGRAM IN CHEMICAL AND PETROLEUM ENGINEERING
CALGARY, ALBERTA
SEPTEMBER, 2014
© David Manzano Angeles 2014
Abstract
The primary objective of this thesis is the evaluation of geomechanical behavior of two
offshore soft sandstone gas reservoirs located in the Gulf of Mexico with a view to quantifying
the geomechanical risk associated with subsidence and compaction.
To meet this objective a 3D Mechanical Earth Model (3D MEM) was built that included:
(1) a reservoir model capable of handling equations governing multiphase flow in porous media
and heat transfer, (2) a geomechanical model that handles equations governing the relationship
between principal stresses, pore pressure, temperature and porosity. Fluid flow models have been
used in the petroleum industry for several decades. On the other hand geomechanical models are
generally considered as newcomers.
Original contributions for the study area include:
(1) Development of correlations between static and dynamic Young’s modulus and
Poisson’s ratio.
(2) Development of correlations that relate the internal friction angle and unconfined
compressive strength to Young’s modulus and porosity.
(3) Quantification of subsidence and compaction.
Basic data for development of items (1) and (2) were provided by sonic-wave velocities
and mechanical laboratory experiments conducted in soft sandstone cores collected in the
reservoirs under consideration.
Item (3) was developed using the 3D MEM. Distribution of rock mechanical properties in
the 3D MEM was developed applying geostatistical data analysis and Sequential Gaussian
Simulation (SGS) methods. The simulation process was used to produce equally probable maps
of mechanical properties.
ii
Computed results using the 3D MEM indicate that average subsidence will be -0.74 m
and average compaction of the upper reservoir -1.37 m under the anticipated production
schedule. This collapse could induce catastrophic damage of subsea production facilities if not
taken into account. Understanding of these displacement processes as presented in this thesis
should help to develop mitigation strategies in order to minimize down to a minimum any risks
associated with well integrity and deep water facilities.
iii
Acknowledgements
Completion of this thesis was possible thanks to those persons who provided support and advice
throughout its development. I express my gratitude to:
My supervisor Dr. Roberto Aguilera for giving me the opportunity to be part of the GFREE
research group. My thanks for his guidance, supervision, teaching and encouragement during the
time we worked together. His knowledge and ability to share ideas have made the completion of
this thesis possible.
My co-supervisor Dr. Gaisoni Nasreldin (Schlumberger) for his supervision, guidance, advice
and recommendations during construction of the Mechanical Earth Model and simulation runs.
He made possible the culmination of this thesis.
Pemex Exploration and Production (PEP) and Conacyt (Mexico) for providing the financial
support that made possible my entire experience as a graduate student in the Department of
Chemical and Petroleum Engineering at the University of Calgary. I thank to Dr. Jorge Alberto
Arévalo Villagrán (PEP) for his support and advice.
My friend and colleague Nicolas Gomez Bustamante, MEng. (Schlumberger, Calgary) for his
support and help during the donation process of Visage software to the University of Calgary.
GFREE members for their continuous support and collaboration. Particularly, thanks to Bruno A.
Lopez Jimenez for his support and advice.
iv
My wife and children for their love, understanding and support during my career development
and for helping me to overcome all difficulties during this time.
v
Dedication
I dedicate this thesis to my family, especially wife Candy Ochoa, to my daughter Isabella
Manzano and my son David Manzano Ochoa. I also dedicate this thesis to my country Mexico.
“We all have dreams. But in order to make dreams come into reality, it takes an awful lot of
determination, dedication, self-discipline, and effort.”
Jesse Owens
vi
Table of Contents
Abstract ............................................................................................................................... ii Acknowledgements ............................................................................................................ iv Dedication .......................................................................................................................... vi Table of Contents .............................................................................................................. vii List of Tables .......................................................................................................................x List of Figures and Illustrations ......................................................................................... xi List of Symbols, Abbreviations and Nomenclature ....................................................... xviii
INTRODUCTION ..................................................................................1 1.1 Justification ................................................................................................................1 1.2 Objective ....................................................................................................................1 1.3 Area of study ..............................................................................................................2
1.3.1 Geologic interpretation and reservoir description .............................................3 1.3.2 Petrophysical model ..........................................................................................5 1.3.3 Gas in place and reserves ..................................................................................7
1.4 History .......................................................................................................................7 1.5 Summary ..................................................................................................................12 1.6 Thesis Chapters ........................................................................................................12
LITERATURE REVIEW ....................................................................14 2.1 Linear Elasticity .......................................................................................................14 2.2 Poroelastic Theory ...................................................................................................17
2.2.1 Effective stress .................................................................................................20 2.2.2 Biot’s coefficient (α) .......................................................................................21
2.3 Rock Mechanics .......................................................................................................24 2.3.1 Relation between acoustic waves and dynamic moduli ..................................24 2.3.2 Static and dynamic moduli ..............................................................................27
2.4 Rock Strength ..........................................................................................................30 2.4.1 Failure criterion ...............................................................................................30 2.4.2 Coulomb criterion ............................................................................................31 2.4.3 Relations between rock strength and physical properties ................................35
2.5 Coupling Reservoir-Geomechanical Models ...........................................................36 2.6 Reservoir Compaction and Subsidence ...................................................................41
MECHANICAL PROPERTIES AND IN-SITU STRESSES .........43 3.1 Static and Dynamic Moduli .....................................................................................43 3.2 Experimental Data ...................................................................................................43 3.3 Mean Effective Stress ..............................................................................................49
3.3.1 Pore pressure ...................................................................................................50 3.3.2 Vertical stress ..................................................................................................50 3.3.3 Minimum horizontal stress ..............................................................................51 3.3.4 Maximum horizontal stress .............................................................................52 3.3.5 Mean effective stress .......................................................................................56
3.4 Static Moduli ............................................................................................................57 3.5 Dynamic Moduli ......................................................................................................59
vii
3.6 Dynamic to Static Young’s Modulus Correlation ...................................................61 3.7 Dynamic and Static Poisson’s Ratio Correlation .....................................................64 3.8 UCS and Internal Friction Angle Estimation ...........................................................66
NUMERICAL RESERVOIR SIMULATION MODEL ...................71 4.1 Structural Grid .........................................................................................................71 4.2 Facies Modeling .......................................................................................................74 4.3 Petrophysical Modeling ...........................................................................................77 4.4 Rock Type Definition ..............................................................................................80 4.5 Reservoir Limits ......................................................................................................81 4.6 Saturation Functions ................................................................................................82
4.6.1 Capillary pressure ............................................................................................82 4.6.2 Relative permeability .......................................................................................83
4.7 Fluid Characterization ..............................................................................................85 4.8 Vertical Flow Performance ......................................................................................89 4.9 Development Strategy ..............................................................................................90 4.10 Reservoir Numerical Simulation Results ...............................................................91
MECHANICAL EARTH MODEL .....................................................93 5.1 Grid ..........................................................................................................................93
5.1.1 The Sideburden ................................................................................................93 5.1.2 The Overburden ...............................................................................................94 5.1.3 The Underburden .............................................................................................96 5.1.4 MEM grid ........................................................................................................97
5.2 Grid Quality Control ................................................................................................98 5.3 Single Material Test .................................................................................................99
5.3.1 Single material test results .............................................................................100 5.4 Mechanical Properties Extrapolation .....................................................................101 5.5 Pre-Production Stress State ....................................................................................108
5.5.1 Minimum horizontal stress ............................................................................108 5.5.2 Stresses orientation ........................................................................................111 5.5.3 Discontinuity modelling ................................................................................113 5.5.4 Pre-production stress state results (elastic run) .............................................116
MECHANICAL EARTH MODEL RESULTS ...................................119 6.1 Stresses State Analysis...........................................................................................119
6.1.1 In-situ stresses orientation .............................................................................119 6.1.2 In-situ stress condition ...................................................................................120
6.2 Sequential Gaussian Simulation ............................................................................124 6.3 Subsidence and Reservoir Compaction .................................................................127
CONCLUSIONS AND RECOMMENDATIONS ........................137 7.1 Conclusions ............................................................................................................137 7.2 Recommendations ..................................................................................................138
REFERENCES ................................................................................................................140
viii
APPENDIX A: GEOSTATISTICAL DATA ANALYSIS .............................................146 A.1 Data Analysis Models ...........................................................................................146 A.2 Distribution of Properties ......................................................................................147
A.2.1 Young’s modulus ..........................................................................................149 A.2.2 Poisson’s ratio ...............................................................................................155 A.2.3 UCS ..............................................................................................................161 A.2.4 Porosity .........................................................................................................167 A.2.5 Bulk density ..................................................................................................173 A.2.6 Friction angle ................................................................................................179
ix
List of Tables
Table 1-1 Lower Miocene petrophysical properties of GML reservoirs. ....................................... 6
Table 1-2 Sample composition of the reservoir fluid. .................................................................. 10
Table 1-3 Well testing Analysis results ........................................................................................ 11
Table 2-1 Elastic moduli (Rider and Kennedy, 2011, pp. 179). ................................................... 25
Table 2-2 Corrections to convert dynamic into static moduli....................................................... 29
Table 3-1 GML experimental data. ............................................................................................... 47
Table 3-2 GMK-1 experimental data. ........................................................................................... 48
Table 3-3 GMK-2 experimental data. ........................................................................................... 49
Table 3-4 Static Young’s moduli (E) and Poisson’s ratios (v) estimated at confining pressure equal to 2100 Psi. .................................................................................................................. 58
Table 4-1 Well logs scale-up process and corresponding settings. .............................................. 78
Table 4-2 Reservoir fluid composition. ........................................................................................ 86
Table 5-1 MEM single material properties. .................................................................................. 99
Table 5-2 Settings for scale-up of properties from well logs ...................................................... 102
Table 5-3 Stiff plate properties. .................................................................................................. 107
Table 5-4 Underburden properties using graded method. .......................................................... 108
Table 5-5 Minimum horizontal stress from leak Off Test .......................................................... 110
Table 6-1 Statistics for distribution of minimum horizontal stress, maximum horizontal stress, vertical stress and pore pressure. .............................................................................. 126
Table 6-2 Univariate Statistics for subsidence and compaction ................................................. 129
Table 6-3 Data required for estimating subsidence and upper reservoir compaction ................. 134
x
List of Figures and Illustrations
Figure 1-1 Location of GML field. ................................................................................................. 2
Figure 1-2 Structural model of GML field defined with the use of seismic data. .......................... 3
Figure 1-3 Conceptual Lower Miocene sedimentary model. .......................................................... 5
Figure 1-4 Reservoir’s rock characteristics. The south-center area is characterized considering GML-1 information whereas the north area is characterized with data from GML-2. The reservoirs of interest are shown by the red boxes .............................................. 6
Figure 1-5 Pressure gradients and gas-water contacts corresponding to lower reservoir (MI-1) and upper reservoir (MI-2). ..................................................................................................... 9
Figure 1-6 Build-up test analysis of well GML-2. ........................................................................ 11
Figure 2-1 Stress and strain ratio. ................................................................................................. 15
Figure 2-2 Longitudinal and transversal strain. ............................................................................ 16
Figure 2-3 Failure plane resulting from shear and normal stresses. ............................................. 30
Figure 2-4 Failure stress................................................................................................................ 32
Figure 2-5 Mohr’s envelope. ......................................................................................................... 34
Figure 2-6 (a) Rock compaction versus effective stress, (b) Rock compaction in Mohr-Coulomb diagram. (Source: Settari, 2002) ........................................................................... 42
Figure 3-1 Triaxial cell (Source: FHWA, 2007) ........................................................................... 44
Figure 3-2 Triaxial test results. (a) Stress versus strain and (b) Mohr-Coulomb circle with failure line. ............................................................................................................................ 46
Figure 3-3 GML-2 bulk density log. ............................................................................................. 51
Figure 3-4 Minimum horizontal stress profile (green line) and leak off test data (red dots). ....... 52
Figure 3-5 Stress polygon. ............................................................................................................ 53
Figure 3-6 GML-2 stress polygon................................................................................................. 56
Figure 3-7 Elastic moduli. (a)Young’s modulus extrapolation and (b) Poisson’s ratio extrapolation. ........................................................................................................................ 58
Figure 3-8 Dynamic moduli. (a) Young’s modulus and shear modulus from wireline logs, (b) Bulk modulus and Poisson’s ratio profiles from wireline logs. ............................................ 59
xi
Figure 3-9 Static and dynamic moduli. (a) Red points represent static Young’s moduli and the blue continuous line corresponds to the dynamic Young’s moduli, (b) Black points represent static Poisson’s ratio values and the red continuous line corresponds to dynamic Poisson’s ratios. ...................................................................................................... 61
Figure 3-10 Young’s modulus correlations. a) Available relationships vs. experimental data, b) Relationships developed in this work vs. experimental data. ........................................... 63
Figure 3-11 Dynamic Young’s moduli in the blue continuous line and converted static Young’s modulus in the yellow continuous line as compared to real experimental data. .... 64
Figure 3-12 Poisson’s ratio in sandstones.(a) Poor linear fit because the experimental data for sandstone GMK-2. (b) Experimental data for sandstone GMK-2 were removed and the linear fit improves ................................................................................................................. 65
Figure 3-13 Conversion of dynamic Poisson’s ratios (red line) to static Poisson’s ratios (yellow line) and its comparison to experimental data. ........................................................ 66
Figure 3-14 Young’s modulus versus UCS. ................................................................................. 68
Figure 3-15 Porosity versus internal friction angle. ...................................................................... 70
Figure 4-1 Reservoir model. (a) Seismic surface and well tops. (b) 3D view of reservoir grid skeleton. ................................................................................................................................ 72
Figure 4-2 Grid quality control. (a) Cell volume attribute and resulting histogram. (b) Cell angle attribute and resulting histogram indicating good geometry in the majority of the cells. ...................................................................................................................................... 73
Figure 4-3 Facies modeling process. ............................................................................................ 75
Figure 4-4 3D Sedimentary facies model. .................................................................................... 76
Figure 4-5 3D lithofacies model. .................................................................................................. 77
Figure 4-6 Scaled-up effective porosity (phie). ............................................................................ 78
Figure 4-7 Porosity and permeability histograms. ........................................................................ 79
Figure 4-8 Rock type definition using Pittman r40. (a) Upper reservoir. (b) Lower reservoir. ... 81
Figure 4-9 Aquifers associated with the upper and lower reservoirs............................................ 82
Figure 4-10 Saturation functions. (a) Leverett J-function and capillary pressure for rock type 1. (b) Relative permeability for wetting and non-wetting phases for rock type 1. ............... 85
Figure 4-11 Phase behaviour diagram before and after the match. .............................................. 87
xii
Figure 4-12 Constant composition expansion experiment at 60.5 °C. Red dots represent experimental data. The lines represent the simulated data.................................................... 88
Figure 4-13 Vertical flow performance table (well gas production rate versus bottom hole pressure) used in well gas production simulation. ................................................................ 89
Figure 4-14 Location of the wells throughout the reservoir under study. .................................... 91
Figure 4-15 Reservoir simulation model results. (a) Original gas in-place histogram built from 300 realizations. (b) Field daily gas production rate and its corresponding cumulative gas production. ................................................................................................... 92
Figure 5-1 Sideburden cells distribution considering geometrical spacing. ................................. 94
Figure 5-2 Surface maps. (a) Seabed surface configuration and (b) Upper Pliocene surface. ..... 95
Figure 5-3 Overburden cells distribution. ..................................................................................... 96
Figure 5-4 Underburden cells distribution. ................................................................................... 97
Figure 5-5 MEM 3D grid. Reservoir and underburden, overburden and sideburden. .................. 97
Figure 5-6 Single material test using the MEM 3D grid. ........................................................... 101
Figure 5-7 Scale-up properties from well logs, well log in solid pink color, scale-up cells in solid green color. (a) Porosity, (b) Young’s modulus, (c) Poisson’s ratio and (d) UCS. ... 103
Figure 5-8 Leak off test developed in well GML-3 at depth of 2996 m. .................................... 109
Figure 5-9 Leak off tests carried out in offshore field under study. There are three different gradients for the minimum horizontal stress: 0.17, 0.18 and 0.19 bar/m ........................... 110
Figure 5-10 Minimum horizontal stress orientation from caliper log and maximum horizontal stress orientation from faults. .............................................................................................. 112
Figure 5-11 Fault extension from lower Miocene to -4500 m. (a) seismic image indicating the fault’s extension in the reservoir, (b) fault’s extension in the geomechanics model. ......... 115
Figure 5-12 Vertical gradient (grey marks), minimum horizontal stress gradient (blue marks) and maximum horizontal stress gradient (orange marks), (a) GML-1 and (b) GML-2. ..... 116
Figure 5-13 1D and 3D comparison for vertical and minimum horizontal stress. (a) and (b) GML-1, (c) and (d) GML-2 ................................................................................................ 118
Figure 6-1 XY plane showing the initial maximum horizontal stress orientation in the reservoir and close-up in blue square. ................................................................................. 120
Figure 6-2 Mohr circles calculated in the 3D MEM and failure envelop at reservoir depth. ..... 121
xiii
Figure 6-3 Photomicrographs of thin sections, 4x and 10x, 100 μm and 300 μm, plane light and polarized light for GML-3 at depth of 3091.3 m. ........................................................ 122
Figure 6-4 Photomicrographs of thin sections, 4x and 10x, 100 μm and 300 μm, plane light and polarized light for GML-2 at depth of 3068.7 m. ........................................................ 123
Figure 6-5 Kinematic classification of deformation bands. (Source: Fossen et al., 2007). ........ 124
Figure 6-6 Two sequential Gaussian simulations realizations. (a) and (b) Young’s modulus. (c) and (d) Poisson’s ratio. .................................................................................................. 125
Figure 6-7 Principal stresses distribution at upper reservoir datum depth. (a) Minimum horizontal stress. (b) Maximum horizontal stress. (c) Vertical stress. (d) Pore pressure. .. 127
Figure 6-8. Time step selection to perform coupling modeling.................................................. 128
Figure 6-9 Univariate histogram using 10 realizations. (a) Subsidence. (b) Upper reservoir compaction. ......................................................................................................................... 130
Figure 6-10 3D MEM rock displacement. (a) Seabed subsidence. (b) Reservoir and overburden compaction. ...................................................................................................... 131
Figure 6-11 Verical displacement vs. reservoir pressure. Subsidence measured at seabed is shown in red. Compaction measured at the upper reservoir depth is shown in blue. ......... 132
Figure 6-12 Schematic ilustrating the main parts of a deepwater subsea tieback production facilities. .............................................................................................................................. 135
Figure A-1 Black squares represent the experimental data, the black line the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction, middle variogram in major direction and lower variogram in minor direction). 149
Figure A-2 3D MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between results from well logs used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. .................................................................................................................................. 150
Figure A-3 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 151
Figure A-4 3D MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 152
Figure A-5 Black squares represent the experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction, middle variogram in major direction and lower variogram in minor direction). 153
xiv
Figure A-6 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 154
Figure A-7 Black squares represent the experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction, middle variogram in major direction and lower variogram in minor direction). 155
Figure A-8 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model.. .................... 156
Figure A-9 Black squares represent the experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction, middle variogram in major direction and lower variogram in minor direction). 157
Figure A-10 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 158
Figure A-11 Black squares represent the experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction, middle variogram in major direction and lower variogram in minor direction). 159
Figure A-12 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 160
Figure A-13 Black squares represent the experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction, middle variogram in major direction and lower variogram in minor direction). 161
Figure A-14 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 162
Figure A-15 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 163
Figure A-16 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 164
Figure A-17 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 165
xv
Figure A-18 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 166
Figure A-19 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 167
Figure A-20 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 168
Figure A-21 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 169
Figure A-22 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 170
Figure A-23 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 171
Figure A-24 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 172
Figure A-25 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 173
Figure A-26 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 174
Figure A-27 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 175
Figure A-28 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 176
Figure A-29 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 177
xvi
Figure A-30 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 178
Figure A-31 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 179
Figure A-32 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 180
Figure A-33 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 181
Figure A-34 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 182
Figure A-35 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction middle variogram in major direction and lower variogram in minor direction). ................ 183
Figure A-36 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the upscaled well log in the cell crossed by the well and the cells in the model. ..................... 184
xvii
List of Symbols, Abbreviations and Nomenclature
Acronyms Definition 1D One-dimensional 2D Two-dimensional 3D Three-dimensional BHP Bottom hole pressure CCE Constant composition expansion experiment CBM Coal bed methane formations DST Drill stem test EOS Cubic equations of state FGPR Field gas production rate FGPT Field gas production total FPR Field reservoir pressure GFREE Integrated geoscience (G), formation
evaluation (F), reservoir drilling, completion and stimulation (R), reservoir engineering (E), economics and externalities (EE)
LVDTs Linear variable differential transformer LOT Leak off test MDT Modular Dynamic Tester MEM Mechanical earth model MI-1 Lower reservoir MI-2 Upper reservoir NTG Net to gross ratio OGIP Original gas in place PVT Pressure-volume-temperature SGS Sequential Gaussian Simulation SSTVD Subsea true vertical depth TOTSTRXX Total stress in x direction TOTSTRYY Total stress in y direction TOTSTRZZ Total stress in z direction UCS Unconfined compressive strength V Velocity XPT Express pressure temperature Nomenclature Definition A Area B(α1, β1) Beta function Cf Fluid compressibility D Reservoir depth E Young’s Modulus Edy Dynamic Young’s modulus Est Static Young’s modulus
xviii
�⃗�𝐹 Restoring force g Gravitational acceleration constant G Shear modulus h Reservoir thickness H Physical constant in Biot’s theory J(Sw) Leverett J function k Proportionality constant K Permeability Kb Bulk modulus Kdy Dynamic bulk modulus Kst Static bulk modulus Kfr Bulk modulus of the skeleton Krnw Non-wetting relative Permeability Krnw* Normalized non-wetting phase relative
permeability Krw Water relative permeability Krw* Normalized water relative permeability Ks Bulk modulus of the solid grain L Length ML Molecular fraction n Number of samples NF Normal Faulting P Compressional Pc Confining pressure Pcap Capillary pressure Pp Pore pressure r Radius ra Rank R Physical constants in Biot’s theory R2 Coefficients of determination RF Reverse faulting s Normal score S Shear SHmax Maximum horizontal stress Shmin Minimum horizontal stress So Shear strength of the rock Srnw Residual non-wetting phase saturation SS Strike- Slip Faulting Sv Vertical stress Sw* Effective saturation Swi Irreducible water saturation U Reservoir subsidence vdy Dynamic Poisson’s ratio vst Static Poisson’s ratio ν Poisson’s ratio var Variance in beta function
xix
vc Poisson’s ratio for cap rock Vclay Clay volume Vp Compressional wave velocity Vs Shear wave velocity Wbo Angle of breakout width. �⃗�𝑥 Displacement distance x� Mean Z Depth of interest Zi Mole fraction Zw Water depth Greek Symbols Definition α Biot’s coefficient α1 Beta function coefficient β Grain compressibility β1 Beta function coefficient γ Shear strain γL Gas relative density δK Changes in permeability δL Change in length δQ Change in wave velocity δr Change in radius δSw Increment of water per unit volume of rock δz Change of stress in z direction ΔC Reservoir compaction ΔP Change in pressure Δtc Compressional slowness Δts Shear slowness Ɛx Axial strain Ɛy Transversal strain Ɛvol Volumetric strain. ϵ Volume increase of soil per unit volume η Quartile point θ Failure plane angle θb Breakout angle λ Pore size distribution. μ Internal friction coefficient ξ 1/ϕ ρb Bulk density ρw water density �̅�𝜌 Mean overburden density σf Stress acting on the fluid, σ Total stress σeff Effective stress σh Stress acting on the fluid
xx
σt Interfacial tension. σx Axial stress σz Stress in z direction τ Shear stress τmax Critical shear stress ϕ Porosity ϕe Effective porosity φ Internal friction angle
Note: This thesis uses mixed units normally employed in the oil and gas industry. The unit kg is
referred as kilogram force. SI units are included in brackets.
Pressure and stress
1 MPa = 10.197 kg/cm2 = 145.037 psi = 10 bar
xxi
Introduction
1.1 Justification
Reservoir engineers historically have paid little attention to mechanical properties of the
rock despite the fact that many physical phenomena occurring in the reservoir’s porous media
including for example sand production, subsidence, compaction drive, water flooding, wellbore
stability, and fracturing can be explained and modeled more easily by considering the effect of
the principal in-situ stresses acting on the mechanical properties of the rock.
In general, better understanding of stresses acting on a reservoir and their effect on
mechanical properties of the rock will improve reservoir characterization and will provide more
accuracy on production forecasts, changes in porosity and permeability, and predictions of
compaction and subsidence.
Analysis of principal stresses acting on the rock and their effect on mechanical properties
require coupling of reservoir flow and rock mechanics models. The integration process is
commonly called Coupled Geomechanical Modeling and can be developed using different
methods which are discussed in more detail later in this thesis.
1.2 Objective
The main objective of this work is to quantitatively evaluate rock compaction and
subsidence in GML Field offshore Mexico, the possibility of sand production, and the reduction
of permeability and porosity during the production stage of the field.
To achieve the objective, three important tasks must be completed,
1) Determine the initial elastic moduli of the rock within the reservoir in the
overburden, underburden and sideburden based on experimental data coming
1
from triaxial tests, wireline logs and correlations to correct the dynamic moduli to
static moduli.
2) Build a coupled reservoir and geomechanical three-dimensional model to simulate
the current stress-strain acting on the porous medium of the reservoir.
3) Compare gas production, pore pressure, pore volume, sand production and
subsidence results with and without the coupled geomechanics modeling. Based
on results, prepare recommendations for proper modeling of the field.
1.3 Area of study
The GML Field is located in the slope and base of the basin floor in the Gulf of Mexico
131 km northwest of Coatzacoalcos, Veracruz, Mexico. The water leg in this area is around 1200
m. The GML Field presents two stacked sandstone reservoirs in the lower Miocene formation
isolated by a cap of shale. Each reservoir is composed by thin layers of soft sandstone
intercalated with thin layers of shale. Figure 1-1 shows the location of GML Field.
Figure 1-1 Location of GML field.
GML
2
1.3.1 Geologic interpretation and reservoir description
The structural model of GML field was defined using seismic interpretation (fast track
post-stack in time) calibrated with data of wells GML-1, GML-2 and GML-3. It is
conceptualized as an elongated anticline with average dimensions of 10.5 km long, 2.1 km wide
and 40 m thick, and major axis orientation in azimuth N 13° W. This structure is limited in the
north by a sealing normal fault and all around by the layer’s dip. It contains 13 principal normal
faults with azimuth N30°E and 80° dip distributed along the structure. The faults are the result of
overburden load as shown in Figure 1-2. The figure does not show all the faults due to their very
low displacement and the scale used in the figure.
Figure 1-2 Structural model of GML field defined with the use of seismic data.
GML-3 GML-1 GML-2
GML-1GML-3
GML-3GML-1
GML-2
3
The Lower Miocene sedimentary model was conceptualized using different sources of
information such as core analysis, drill cuttings, wireline logs, seismic stratigraphy and
paleontological analysis. The data analysis indicated that these reservoirs present facies of levee
channel within turbidite deposits corresponding to bathyal depositional environment. The
sedimentary structures are notable because of the presence of flame structures and convolute
bedding of fine sands in between soft-sediment deformed structures, parallel lamination and
cross-planar stratification as presented in Figure 1-3. The low amplitude seismic attribute
allowed identifying a channel with orientation NE-SW associated with a levee channel, which is
the source of the sand deposits identified during drilling of wells in both reservoirs.
The stratigraphic model was conceptualized dividing the reservoirs in two areal zones;
the south-centre and north zones. The former utilized information gathered during drilling and
completion operations of GML-1 and GML-3, which helped to identify the presence of
intercalated laminar layers of sandstones and shales in both reservoirs. On the other hand, the
north zone used information gathered during drilling and completion operations of appraisal well
GML-2, which allowed distinguishing massive sandstone in the upper reservoir and laminar
layers of sandstone intercalated with layers of shale in the lower reservoir.
The seal rock in both reservoirs is comprised basically of shale with an average thickness
is 30 m. The isotopic analysis of gas samples from both reservoirs indicated that the source rock
is most likely the Upper Jurassic in the area.
4
Figure 1-3 Conceptual Lower Miocene sedimentary model.
1.3.2 Petrophysical model
In general, the lithology of the rock in both reservoirs is constituted by siliciclastic
sediments consisting of quartz, potassium feldspars, sodium rich plagioclase and igneous clasts
cemented with calcite and clay. Both reservoirs present vertical heterogeneity mainly due to
strong variation in sandstone’s texture, which is related to grain size, clay content, degree of
consolidation and classification. This stratification results in changes in the porous system and
storage quality in the rock. Core samples and wireline logs analyses indicated that both
reservoirs in the south-center area have laminated characteristics. In the north area the lower
GML-1
5
reservoir presents the same characteristic as the ones in the south-center area, but the upper
reservoir has a massive structure. Figure 1-4 shows each of the wells considered for analysis and
their corresponding rock characteristics. Table 1-1 presents the main petrophysical
characteristics by reservoir and areal zone.
Table 1-1 Lower Miocene petrophysical properties of GML reservoirs.
Rock Property Upper Reservoir Center North
Lower Reservoir Center North
Porosity 20 % 19.5% 19.6 % 18 % Permeability 35 mD 37.5 mD 25.5 mD 33 mD Clay volume 13% 14 % 16 % 17 % Water saturation 38% 31 % 46 % 43 % Net thickness 42 m 60 m 14 m 11 m
Figure 1-4 Reservoir’s rock characteristics. The south-center area is characterized considering GML-1 information whereas the north area is characterized with data from
GML-2. The reservoirs of interest are shown by the red boxes
Lam
inat
edL
amin
ated
Lam
inat
edL
amin
ated
Mas
sive
GML-1 GML-2
6
1.3.3 Gas in place and reserves
Based on the number of wells completed in this field the original gas in-place (OGIP) and
2P reserves (proven reserves + probable reserves) have been estimated three times during the
development life of the reservoirs. The first estimate was performed after completion of well
GML-1 in March 2007. The original gas in-place was estimated at 1,733 MMMscf and the 2P
reserves at 934 MMMscf. The second estimation was performed in July 2010, after completion
of well GML-2 (2008), which was the first appraisal well located in the north of the field. The
OGIP was revised to 1,128 MMMscf and 2P reserves were 866 MMMscf. The last
reclassification was done in 2013 after completion of well GML-3, which is the second appraisal
located in the south of the field. The third estimate led to an OGIP equal to 1,298 MMMscf while
2P were 922 MMMscf. According to the last reclassification, the whole reserves in the field have
been categorized as 2P.
1.4 History
Well GML-1 was drilled to a depth of 3818 m between July 2006 and March 2007. The
seabed was located at a depth of 870 m. The found stratigraphic column comprises recent
Pleistocene and lower Miocene. The completion stage included the perforation of four zones.
The lower zone was dry, two intervals in the middle zone produced natural gas and the upper
zone produced 100% water. The upper part in the middle zone is named the upper reservoir and
was completed in interval 3035 - 3127 m. Similarly, the lower part in the middle zone matches
with the lower reservoir and was completed in interval of 3174 to 3212 m. Each interval was
evaluated with the use of PVT bottom-hole samples, surface chromatography and flow testing.
The PVT and chromatography analyses indicated the presence of dry gas with 96% mole fraction
7
of methane. Flow testing conducted through a 5/8” choke size resulted in gas rates equal to 24
MMscfd and 36 MMscfd for the lower and upper reservoirs, respectively.
Well GML-2 was perforated in July 2010 at interval 3075-3100 m. Information was
gathered by different means such as downhole fluid analysis using the modular formation
dynamic tester (MDT), PVT bottom-hole samples, surface chromatography, well testing and wall
cores during the completion process. A total of 58 pressure and temperature data points were
measured using mini DST, XPT and DST to determine pressure and temperature gradients.
Pressure gradients in the gas zone and water zone were 0.0232 kg/cm2/m (2.35 KPa/m) and
0.134 kg/cm2/m (13.1 KPa/m), respectively. This information was used to establish both top and
base of each interval as follows,
1) Upper reservoir from 3048 to 3119 m.
2) Lower reservoir from 3186 to 3214 m.
Bottom-hole fluid samples and dynamic well testers allowed identifying the gas-water
contact for both reservoirs. The lower reservoir (MI-1) gas-water contact was located at a depth
of 3189 m SSTVD while gas-water contact in the upper reservoir (MI-2) was located at a depth
equal to 3098 m SSTVD. Figure 1-5 shows pressure gradients and gas-water contacts for both
reservoirs.
8
Figure 1-5 Pressure gradients and gas-water contacts corresponding to lower reservoir (MI-1) and upper reservoir (MI-2).
Chromatographic analysis was performed on both bottom-hole and recombined samples
collected in the surface. Both analyses indicated that the fluid is dry gas. The composition of the
fluid is shown in Table 1-2 and indicates 0.0% mole of H2S and 0.03-0.10% mole of carbon
dioxide.
However, the fluid can also be classified as a poor gas condensate. In fact, the dew point
pressure is 383 kg/cm2 (37.55 MPa), relative gas density is equal to 0.59 and gas-oil ratio is
equal to 3.9 bbl/MMscf at standard conditions. The dew pressure is larger than the initial
reservoir pressure, which is an indication of a gas condensate reservoir. However, strong
condensation is not expected in the reservoir since the liquid content is low.
Lakach-1 (DST)
Lakach-2DL (XPT)
miniDSt
Flow test
Sample recovered
Pressure, kg/cm2 (1kg/cm2=98.06 KPa)
Dep
th, m
9
Table 1-2 Sample composition of the reservoir fluid.
Component Bottom-hole sample composition
GML-2 GML-1 N2 1.03 0.84
CO2 0.03 0.10 C1 95.07 94.81 C2 1.97 1.90 C3 0.72 0.74 C4 0.38 0.20 C5 0.16 0.18 C6 0.17 0.06 C7 0.14 0.07 C8 0.34 1.11
Total 100.00 100.00
Three well-testing analyses were carried out for these reservoirs. Two of them were
performed in well GML-1 and the other one in well GML-2.
The first test, conducted in December 2006, covered interval 3173-3193 m. The second
test was performed in January 2007 and covered interval 3047-3095 m. The last test was
performed in July 2010 at interval 3075-3100 m. The three tests consisted of drawdowns through
chokes of 3/8, 5/8 and ½-inches along with build-ups, which lasted around 100 h. The model
used to match the real pressure data indicated a radial homogeneous reservoir with partial
penetration and boundary effects possibly representative of a facies change. Figure 1-6 shows
the build-up test for GML-2. Table 1-3 presents the most important results obtained from the
analysis.
10
Table 1-3 Well testing Analysis results
GML-1 GML-1 GML-2 Interval 3173-3193 m TVD 3047-3095 m TVD 3075-3100 m TVD
Model Radial Homogeneous with boundary
Radial Homogeneous Radial Homogeneous
Initial pressure @ datum 37.04 MPa 36.08 MPa 36.11 MPa
Gas rate 24.8 MMSCF @ 5/8” 29.5 MMSCF @ 5/8” 28.68 MMSCF @ 11/16”
φe 19% 23% 21% Total skin -3 15.7 50.5 Net depth 15 m 36 m 51 m Permeability 3.5 mD 24.7 mD 31.6 mD Research radius 379 m 676 m
ΔP 8.5 MPa 2.8 MPa 5.3 MPa
Figure 1-6 Build-up test analysis of well GML-2.
High Δ Pressure indicating high skin
Possible effect of partial penetration
Radial Flow
m(p
)-m(p
@dt
=0) a
nd d
eriv
ativ
e [p
si2 /cp
]
0.01 0.1 1.0 10 1001E+6
1E+7
1E+8
Time [hr]
11
1.5 Summary
GML Field includes two stacked reservoirs of Lower Miocene age. Both of them are
composed by sand layers intercalated with shale layers. The reservoir’s structure is an elongated
anticline with an average thickness of 30 m and an area of 20 km2. Their average porosity is
19.3%, average initial water saturation is 39.5% and average permeability is 32.7 mD. The total
volume of original gas-in-place is 1,298 MMMscf and the 2P reserves are in the order of 922
MMMscf. Three wells have been drilled in this field; an exploration well and two appraisal
wells. Well testing data and basic geophysical well logs such as density, shear and compressional
sonic waves velocities are available for both reservoirs. These logs are critical for estimation of
geomechanical properties as explained later in this work.
1.6 Thesis Chapters
This thesis has been structured in seven chapters as follows:
Chapter 1 discusses the main objectives of this work and presents a general description
of the reservoir. The description includes location, geologic summary petrophysical
characteristics, original gas in-situ, reserves and production history.
Chapter 2 presents a literature review of the most important concepts and theories
related to linear elasticity, poroelastic theory, effective stress, static and dynamic elastic
modulus, failure theory, geomechanical modelling and coupling methods.
Chapter 3 describes the methodology followed to develop new empirical correlations
that allow estimates of Young’s modulus, Poisson’s Ratio, uncompressive rock strength and
friction angle of rocks in the study area based on laboratory tests.
12
Chapter 4 shows the applied methodology for numerical reservoir simulation using
stochastic geostatistical models for estimating average rock characteristics, facies, petrophysical
properties and fluid properties.
Chapter 5 describes the process to create the earth mechanical model (MEM); grid
building and validation test, scale-up properties using well logs and empirical correlations, and
stochastic geostatistical simulation to estimate reservoir properties. Also included is a discussion
on elastic MEM stress-state representativeness and one-way coupling simulation to verify the
current stress-strain state in the reservoir model.
Chapter 6 discusses the principal results of two-way stress-flow coupled modeling,
highlighting the importance of permeability and porosity reduction due to stress changes.
Production forecasts by the reservoir model and the full field geomechanical model are
compared. Subsidence results and their implications during the field’s production life are
discussed in this chapter.
Chapter 7 states the main conclusions derived from this work and presents
recommendations for future studies.
13
Literature Review
As previously mentioned, the main objective of this thesis is to analyze the behaviour of
an offshore gas reservoir using a reservoir numerical simulation model coupled with
geomechanics. This is done to improve the accuracy on the forecast estimation of reservoir
production, pore pressure, in-situ stresses and mechanical properties.
This chapter presents a literature review on topics related to this thesis such as linear
elasticity theory, poroelastic theory, empirical relations used for determining mechanical
properties of rocks as well as available techniques for coupling fluid flow and geomechanics
models.
2.1 Linear Elasticity
The fundamentals of rock mechanics reside on the theory of linear elasticity also known
as Hooke’s law, which states that the deformation experienced by an elastic material is linearly
proportional to the stress applied. Thus, the force to restore the material to equilibrium is
proportional to the displacement of the material from its original position expressed as Eq. 2-1
(Davis, 2002).
�⃗�𝐹 = −k�⃗�𝑥 Eq. 2-1
Where �⃗�𝐹 is the restoring force, k is the proportionality constant and �⃗�𝑥 is the displacement
distance. The minus sign indicates that the force acts in the opposite direction of the
displacement.
The linear elasticity and isotropic model (Hooke’s law) is the most simple and also the
most practical way to model and estimate mechanical properties of a material. However, its
application is only valuable as a first estimation due to its simplicity.
14
If a force �⃗�𝐹 is applied at the end of a long thin bar with cross section area A, the length L
will change by δL as shown in Figure 2-1.
Figure 2-1 Stress and strain ratio.
The proportionally constant defined by the ratio of the axial stress σx = �⃗�𝐹/A to the axial
strain Ɛx = δL/L is known as Young’s Modulus (E) as defined in Eq. 2-2. Young’s Modulus
represents the stiffness of the material, i.e., the resistance of the material to be compacted under
uniaxial stress (Fjaer, 2008). In addition, Zoback (2006) pointed out that this measurement of
stiffness must be developed under unconfined conditions for linear elasticity purposes.
𝐸𝐸 =𝜎𝜎𝑥𝑥𝜀𝜀𝑥𝑥
Eq. 2-2
When an axial stress is applied on a rod as illustrated above, the body suffers either
compression or extension (Ɛx) depending on the direction of the stress (this work assumes that
compression is caused by positive stress) and either expansion or contraction (Ɛy) perpendicular
to the axial stress. As shown in Figure 2-2, compression occurs in the direction of the stress
whereas expansion is arising perpendicular to the axial stress. The ratio of lateral to longitudinal
strain is known as Poisson’s ratio (ν) as expressed in Eq. 2-3.
𝑣𝑣 = 𝜀𝜀𝑦𝑦𝜀𝜀𝑥𝑥
Eq. 2-3
�⃗�𝐹 L
δL
L
A σx = �⃗�𝐹 A
Ɛx = δL
15
Figure 2-2 Longitudinal and transversal strain. The stress and strain parameters previously mentioned for the case of a 1D model are
known as normal stress and normal strain, respectively. However, when the 1D model is
extended to 2D or 3D, shear stress and shear strain have to be considered for angular distortion.
Biot (1941) considered a small 3D (Cartesian coordinates) element of porous rock
saturated with water as homogeneous and infinitesimal scale compared with the real scale of the
phenomena and developed the expressions to relate strain to stress presented in Eq. 2-4 to Eq.
2-9.
𝜀𝜀𝑥𝑥 =𝜎𝜎𝑥𝑥𝐸𝐸−𝑣𝑣𝐸𝐸�𝜎𝜎𝑦𝑦 + 𝜎𝜎𝑧𝑧� +
𝜎𝜎ℎ3𝐻𝐻
Eq. 2-4
𝜀𝜀𝑦𝑦 =𝜎𝜎𝑦𝑦𝐸𝐸−𝑣𝑣𝐸𝐸
(𝜎𝜎𝑥𝑥 + 𝜎𝜎𝑧𝑧) +𝜎𝜎ℎ3𝐻𝐻
Eq. 2-5
𝜀𝜀𝑧𝑧 =𝜎𝜎𝑧𝑧𝐸𝐸−𝑣𝑣𝐸𝐸�𝜎𝜎𝑦𝑦 + 𝜎𝜎𝑥𝑥� +
𝜎𝜎ℎ3𝐻𝐻
Eq. 2-6
𝛾𝛾𝑥𝑥 =𝜏𝜏𝑥𝑥𝐺𝐺
Eq. 2-7
𝛾𝛾𝑦𝑦 =𝜏𝜏𝑦𝑦𝐺𝐺
Eq. 2-8
𝛾𝛾𝑧𝑧 =𝜏𝜏𝑧𝑧𝐺𝐺
Eq. 2-9
Where γ is the shear strain in a perpendicular direction to the normal strain, τ is the shear stress
in the perpendicular direction to the normal stress, σh is the stress acting in the fluid, H is a
physical constant that takes into account the effect of water pressure and G is the shear modulus,
�⃗�𝐹
δL
Lδrr
LƐx = δL
rƐy = δr
16
which represents the resistance of the rock to be deformed by a shear stress. The relationship
between shear modulus and Young’s modulus and Poisson’s ratio is given by Eq. 2-10:
𝐺𝐺 =𝐸𝐸
2(1 + 𝑣𝑣) Eq. 2-10
The bulk modulus (Kb) is another important elastic property of the rock to be considered
for geomechanical analysis. The bulk modulus represents the resistance of the rock to be
compressed by hydrostatic stresses acting in all directions. It is denoted by the following Eq.
2-11 (Fjaer, 2008),
𝐾𝐾𝑏𝑏 =𝜎𝜎ℎ𝜀𝜀𝑣𝑣𝑣𝑣𝑣𝑣
Eq. 2-11
Where σh is the hydrostatic stress and Ɛvol the volumetric strain.
Due to the linear relationship between the different modulus, it is possible calculate any
elastic property of the rock as function of two other parameters. Several authors have proposed
different relationships to calculate the elastic parameters E, v, G and Kb. Davis and Selvadurai
(2005) proposed 30 different relationships, Fjaer (2008) established 21 relationships, Zoback
(2002) presented more than 30 relationships and Jaeger et al. (2007) proposed 9 different
equations to calculate elastic properties and a complete set of equations to calculate stress and
strain for specific cases of cylindrical and polar coordinate systems either in two or three
dimensions. However, special care is needed when applying these relations on rocks that do not
exhibit linear stress-strain relationship.
2.2 Poroelastic Theory
As mentioned above the theory of linear elasticity presents limitations because it treats
the rock as homogeneous, solid, isotropic and with a linear stress-strain relationship. Due to
these limitations the linear elastic theory is not applicable to materials conformed by an elastic
17
skeleton containing interconnected pores saturated with compressible fluid that exhibits an
elastic behaviour depending on both solid stress and fluid pressure. In their study Detournay and
Chen (1993) stated that the presence of a mobile fluid phase in the porous system alters the
mechanical behaviour of the rock, so that any compression of the rock increases the pore
pressure, which in turn induces dilation of the rock.
Terzaghi’s 1923 study (cited in Zoback, 2002) took into account the effect of fluid in the
pore system over the deformation of soils. Considering a constant pressure load acting over a
column of porous rock saturated with water and disallowing for lateral expansion, Terzaghi
demonstrated that the behaviour of a porous rock is mainly governed by the pressure difference
between the confining pressure and the pore pressure. However, Terzagui’s work has some
limitations: the work was conducted on 1D models and on rocks with incompressible skeleton.
Biot’s theory (1941) was developed on a 3D model considering the following
characteristics,
1. Porous system is filled with free fluid that supports part of the total stresses acting on the
rock,
2. A coherent solid skeleton supports the remaining part of the total stress, so single stress is
constituted by two specific elements: the hydrostatic pressure and the mean skeleton’s
stress.
Some important assumptions are isotropic material, reversible stress-strain relationship,
linear stress-strain behaviour, incompressible fluid, and fluid flow governed by Darcy’s Law.
Detournay and Chen (1993) highlighted an additional limitation on time scale since the pressure
in a single pore must be equilibrated with the pressure of the surrounding pores (deformation-
18
diffusion process), which does not represent the basis of Biot’s theory. However, the complete
justification of Biot’s Theory resides on the quasi-static process.
Eq. 2-12 to Eq. 2-21 are approximations for estimating properties of the rock saturated
with fluid, under strain equilibrium, isotropic soil and small strain conditions,
𝜎𝜎𝑥𝑥 = 2G �𝜀𝜀𝑥𝑥 +𝑣𝑣𝑣𝑣
1 − 𝑣𝑣� − 𝛼𝛼𝜎𝜎ℎ Eq. 2-12
𝜎𝜎𝑦𝑦 = 2G �𝜀𝜀𝑦𝑦 +𝑣𝑣𝑣𝑣
1 − 𝑣𝑣� − 𝛼𝛼𝜎𝜎ℎ Eq. 2-13
𝜎𝜎𝑧𝑧 = 2G �𝜀𝜀𝑧𝑧 +𝑣𝑣𝑣𝑣
1 − 𝑣𝑣� − 𝛼𝛼𝜎𝜎ℎ Eq. 2-14
𝜏𝜏𝑥𝑥 = G𝛾𝛾𝑥𝑥 Eq. 2-15
𝜏𝜏𝑦𝑦 = G𝛾𝛾𝑦𝑦 Eq. 2-16
𝜏𝜏𝑧𝑧 = G𝛾𝛾𝑧𝑧 Eq. 2-17
𝛿𝛿𝛿𝛿𝑤𝑤 = αϵ +𝜎𝜎ℎ𝑞𝑞
Eq. 2-18
𝛼𝛼 =2(1 + 𝑣𝑣)
3(1 − 2𝑣𝑣)𝐺𝐺𝐻𝐻
Eq. 2-19
𝑞𝑞 = R −𝐻𝐻𝛼𝛼
Eq. 2-20
∈=3(1 − 2𝑣𝑣)
𝐸𝐸𝜎𝜎1 +
𝜎𝜎ℎ𝐻𝐻
Eq. 2-21
Where σx represents the stress acting on the skeleton, σh is the stress acting on the fluid, ϵ is the
volume increase of soil per unit volume, α is the Biot’s coefficient, δSw is the increment of water
per unit volume of rock and H and R are physical constants.
According to Biot and Willis (1957), the Biot’s coefficient mentioned above can be
determined using laboratory sample tests such as the jacketed and drained test. Both tests require
samples saturated with fluid and allow measuring the bulk modulus of the skeleton (Kfr) since
pore pressure is held constant while an external hydrostatic pressure is applied in such a way that
19
the stress is supported by the solid frame. On the other hand, the unjacketed test allows for
equilibrium between pore pressure and hydrostatic pressure. As a result, the stress is uniform and
Ɛvol to t= Ɛvol por e = Ɛvol soli d and the bulk modulus of the solid grain (Ks) can be estimated using
Eq. 2-22. Furthermore, the aforementioned laboratory tests are used to determine porosity (ϕ),
shear modules (G) and fluid compressibility (Cf). Also the five variables defined by Biot’s
theory are calculated from data obtained from these laboratory measurements. The mathematical
expression to calculate the α coefficient as a function of the effective stress (σeff), total external
stress (σ) and pore pressure (Pp) is given as follows,
𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 =𝑃𝑃𝑝𝑝𝐾𝐾𝑠𝑠
Eq. 2-22
𝛼𝛼 = 1 −𝐾𝐾𝑓𝑓𝑓𝑓𝐾𝐾𝑠𝑠
Eq. 2-23
𝜎𝜎𝑒𝑒𝑓𝑓𝑓𝑓 = 𝜎𝜎 − 𝛼𝛼𝑃𝑃𝑝𝑝 Eq. 2-24
2.2.1 Effective stress
Different authors have analyzed the effective stress (σeff). For instance, Terzaghi (1949)
established that σeff acting over a boundary of an element is equal to the difference between the
vertical stress (σz) and the pore pressure (Pp) and is expressed in Eq. 2-25. Nevertheless, this
expression is only valid for 1D cases. Biot (1941) developed the equation as represented in Eq.
2-24., which calculates σeff as a function of σ, Pp and Biot’s coefficient (α)
𝜎𝜎𝑒𝑒𝑓𝑓𝑓𝑓 = 𝜎𝜎𝑧𝑧 − 𝑃𝑃𝑝𝑝 Eq. 2-25
The expression shown in Eq. 2-24 indicates that the total external stress acting over a
boundary of an element is composed by σeff and αPp, where the fluid in the porous medium
supports a fraction of the pore pressure (αPp) and the skeleton carries the effective stress.
20
Meanwhile, inner stresses in the skeleton absorb the residual pore pressure (1-α) Pp (Fjaer,
2008).
2.2.2 Biot’s coefficient (α)
Biot’s coefficient has two physical explanations: the first one states that Biot’s coefficient
is the result of an elastic potential energy present in the rock; the second one indicates that it
represents the fraction of pore pressure (Pp) that produces a strain equal to the total stress (σ). In
any case, Eq. 2-23 indicates that α should take values between zero and one because the bulk
modulus of solid grains is always greater than the bulk modulus of the skeleton (Biot and Willis,
1957).
Based on measurements of acoustic velocities and elastic properties on sandstone
samples Ojala and Fjaer (2007) developed an expression to calculate α as given by Eq. 2-26. In
their measurements, Ojala and Fjaer (2007) considered stress and strain in both radial and
vertical directions, as well as P and S wave velocities in the aforementioned directions (δQ).
Their results indicated that α varies from 0.60 to 0.83 when elastic methods are used, 0.10 to
0.88 for radial P-waves, 0.57 to 0.86 for axial P-waves, and 0.75 to 1.12 for S-waves.
α = 1 −�δQδPp
�𝜎𝜎eff
� δQδσeff
�Pp
Eq. 2-26
Seismic velocity response of underground rocks is a function of overburden and pore
pressure, which have an effect on the burial stress and consequently on the response of acoustic
seismic velocities of the rock. Thus, a relationship between seismic velocity and effective stress
must be defined. Eq. 2-27 allows estimation of α as a function of compressional wave velocity
(Vp), shear wave velocity (Vs), bulk modulus of the solids (Ks), bulk density (ρb) and effective
21
stress determined from seismic velocity and the difference between confining pressure and pore
pressure (ΔP). This equation has been applied to the Carnarvon basin formation to determine
Biot’s coefficient using P and S -wave velocity from cores. The value was found to be
approximately 0.9 (Siggins et al., 2004)
𝛼𝛼 = 1 − 𝜌𝜌𝑏𝑏�3 �𝑉𝑉𝑝𝑝(∆𝑃𝑃)�
2− 4�𝑉𝑉𝑠𝑠(∆𝑃𝑃)�
2�𝑃𝑃𝑝𝑝
3𝐾𝐾𝑠𝑠
Eq. 2-27
Alam and Fabricius (2012) stated that a more precise value of Biot’s coefficient can be
obtained measuring the bulk compressibility of samples under uniaxial strain conditions (more
representative in hydrocarbon reservoirs) and not only the bulk compressibility measured under
hydrostatic stress conditions as traditionally done. As a result, they developed a new expression
to calculate Biot’s coefficient in each main stress direction. The bulk strain is composed of both
radial and axial strain, where the radial strain is the result of tension stress while the axial strain
originates from compression.
When reservoir fluids are extracted, pore pressure decreases and net stress increases
leading to deformation of the grain structure, which in turn reduces porosity. This modification
develops variations in Biot’s coefficient in different directions. Using seismic waves’ velocity
Alam and Fabricius (2012) determined the effective stress and the change in volume due to
compression of a pack. Then, by application of Eq. 2-28, they found Biot’s coefficient to range
between 0.6 and 1 for Austin chalk (Texas) and North Sea chalk reservoirs.
𝛼𝛼 = 1 −
�𝜕𝜕𝜀𝜀𝑥𝑥𝜕𝜕𝑝𝑝𝑝𝑝�𝜎𝜎𝑒𝑒𝑒𝑒𝑒𝑒
�𝜕𝜕𝜀𝜀𝑦𝑦𝜕𝜕𝜎𝜎𝑒𝑒𝑓𝑓𝑓𝑓
�𝑝𝑝𝑝𝑝
Eq. 2-28
22
Vincke et al. (1998) investigated Biot coefficient for shale rocks in an elastic domain. In
their work, they managed Biot coefficient either as a tensor (αij) for the case of anisotropic rocks
or as a scalar for isotropic rocks. They also presented a mathematical expression to calculate
Biot’s coefficient for anisotropic rocks under overburden stress. They assumed that the
behaviour of the stress-strain in the parallel plane to the stratification dip is the same in both
directions of the orthogonal axes lying on this plane. They developed laboratory experiments on
vertical and horizontal plugs and applied vertical stress to quantify the plug’s vertical strain.
Then, under constant overburden pressure, they increased the pore pressure and measured the
vertical strain. Finally, their results indicated that Biot’s coefficient is a function of the different
stresses acting on the sample. Their Biot’s coefficients results ranged from 0.6 to 0.8 with
vertical stresses of 14MPa and 35 MPa respectively.
Qiao et al. (2011) estimated Biot’s coefficient for the Nikanassin tight gas formation
based on permeability measurements of samples collected in outcrops of the tight gas Nikanassin
formation in the Western Canada Sedimentary Basin. The composition of the samples included
quartz, shale and feldspar with grain size varying from fine to coarse.
The samples were tested and absolute permeability values were determined under various
conditions. Qiao et al. (2011) applied modified methods from the relationship proposed by Ojala
and Fjaer (2007) in order to determine Biot’s coefficient as a function of permeability. The
modified expression utilized by Qiao (2011) is as follows,
𝛼𝛼 = −�𝛿𝛿𝐾𝐾𝑃𝑃𝑃𝑃𝛿𝛿𝐾𝐾𝑃𝑃𝑝𝑝
� Eq. 2-29
Where δKPc represents the changes in permeability when the confining pressure is held constant
and the pore pressure is increased, and δKPp considers the changes in permeability when the pore
23
pressure is held constant and the confining pressure is increased. Qiao et al. (2011) applied Eq.
2-29 to estimate Biot’s coefficient for the Nikanassin Formation. They found that average values
are 0.701 in vertical direction and 0.174 in horizontal direction.
According to Zoback (2006), the equation for estimating effective stress presented by
Biot (1941) (Eq. 2-24) does not work well for permeability. Thus Zoback indicates that the
equation should consider a modification on the parameter α in order to be applicable. This
modification leads to α values greater than 1.0 for sandstones with high content of clay mineral.
This α alteration is due to the dependency of permeability on clay content, which implies that α
is less sensible to the confining pressure than to the pore pressure. However Kwon et al. (2002)
indicate that this alteration is not applicable for shale with high content of clay.
In practice, Biot’s coefficient is a function of many factors such as grain mineralogy, pore
geometry, fluid saturation, confining pressure, clay content, fluid compressibility and anisotropy
of the rock. Thus, the relationship to estimate Biot’s coefficient should be based on all the
available information.
2.3 Rock Mechanics
2.3.1 Relation between acoustic waves and dynamic moduli
Seismic acoustic waves are disturbances generated at a specific point on the earth that
transfer energy to the subsurface and travel long distances. The velocity of propagation of the
waves is related to the properties of the rock in two ways; directly based on density of the rock
and its stiffness, and indirectly based on other parameters such as porosity, lithology, pore
pressure, and elastic moduli. Therefore, acoustic waves provide specific information about the
medium of propagation.
24
There are two types of waves involved in water saturated rocks. The first ones are
compression waves, which are also termed as primary waves (P-waves) in which the particles
move in parallel direction to the propagation. The second ones are shear waves (S-waves) in
which the particles move in perpendicular direction to the waves’ propagation. In addition, P-
waves are the only types of waves that can travel in liquids and always arrive first at the
receivers. According to Fjaer et al. (2008), the elastic behaviour of a porous rock is substantially
affected by the saturation fluid. An unconsolidated sandstone will present lower P-wave velocity
when it is saturated with gas than when it is saturated with water. It is because the added
resistance against compression provided by the fluid in the porous medium. In consolidated
sandstones the influence of the pore fluid is quite less. On the other hand, it is considered that S-
waves are not affected by pore saturating fluids.
The equations presented in Table 2-1 are used to calculate the dynamic elastic moduli
when density, sonic compressional slowness and sonic shear slowness are known.
Table 2-1 Elastic moduli (Rider and Kennedy, 2011, pp. 179).
Symbol Elastic Moduli Equation E Young’s modulus, psi 2𝐺𝐺(1 + 𝑣𝑣)
v Poisson’s ratio 0.5(∆𝑡𝑡𝑠𝑠/∆𝑡𝑡𝑃𝑃)2 − 1(∆𝑡𝑡𝑠𝑠/∆𝑡𝑡𝑃𝑃)2 − 1
G Shear modulus, psi 1.34𝑥𝑥1010𝜌𝜌𝑏𝑏∆𝑡𝑡𝑠𝑠2
Kb Bulk modulus, psi 1.34𝑥𝑥1010𝜌𝜌𝑏𝑏 �1∆𝑡𝑡𝑃𝑃2
−4
3∆𝑡𝑡𝑠𝑠2�
C Bulk compressibility, psi-1 𝐾𝐾𝑏𝑏−1
Where:
Δtc = Compressional slowness, μs/ft.
Δts = Shear slowness, μs/ft.
25
ρb = Bulk density, g/cm3.
Artificial acoustic seismic waves are generated underground to gather information about
the elastic properties and the structural configuration of the rock. These seismic waves can be
created in two different locations: at the surface of the earth, and downhole at the well. Acoustic
waves at borehole are short-length and high-frequency; for example, any underground wave-
length of 0.5 m and wave frequency of 10 kHz produce a wave velocity of 5000 m/s. In contrast,
subsurface seismic waves are large-length and low-frequency (Close et al., 2009).
Several authors have proposed alternate methods for estimating dynamic elastic moduli
of the rock using sub seismic acoustic information. For instance, Gassmann (cited in White,
1991) derived some analytical expressions to calculate elastic parameters of a saturated porous
medium by using seismic waves and sample measurements of both fluid and skeleton. However,
he neglected the effect of motion solid-fluid on the seismic wave spread. In 2007, Mullen et al.
suggested a methodology to calculate P-waves and S-waves velocities using petrophysical
parameters from wireline logs when sonic parameters of wireline logs are not available. In 2010,
Banik et al. developed a method for estimating the dynamic moduli using elastic impedance from
subsurface seismic data for unconventional basins. Zhang et al. (2010) developed a methodology
similar to Mullen et al. (2007), but they added 3-D seismic data to populate a geomechanical
model. Trudeng et al. (2014) used a 3-D seismic inversion model to populate the mechanical
properties of the rock in a 3-D geomechanical model incorporating the uncertainties of a sub
seismic model.
In 1934, Conrad Schlumberger proposed a technique to measure sonic wave velocity at
boreholes. In this technique, the sonic log tool measures the capacity of a rock to propagate
acoustic waves. It does so by actually measuring the time that a pulse of sound takes to travel a
26
specific distance. It is the inverse of formation slowness (1/Δt). Since then, sonic wireline logs
have become one of the most important tools for measuring sonic wave velocity (cited in Rider
and Kennedy, 2011). Nowadays, sonic tools can measure compressional and shear slowness
along with Stoneley waves. The use of sonic logs and the application of equations in Table 2-1
allow geoscientists to easily estimate the elastic moduli of the rock.
2.3.2 Static and dynamic moduli
Although sonic waves offer an alternate way to estimate mechanical properties of the
rock, the fact is that generally there is a significant difference between mechanical properties
assessed using the expressions presented in Table 1.1 and mechanical properties determined by
stress and strain measurements from uniaxial or triaxial tests and the application of equations Eq.
2-2 or Eq. 2-30. The elastic properties obtained from sonic wave velocity and rock density are
known as dynamic moduli while those estimated from stress and strain relationships are called
static moduli.
The reasons for differences between static and dynamic moduli have been studied by
several authors. For example, Fjaer (1999), based on triaxial test measurements of stress-strain,
P-waves and S-waves, stated that as a result of changes in stress and strain, the continuous
process of failure is one of the most important reasons for the difference between static and
dynamic bulk modulus in weak sandstones. Montmayeur and Graves (1985) studied the variation
of fluid saturation during testing conditions and found that this is a factor that contributes for the
difference between static and dynamic moduli, especially because static moduli are best defined
with differential variations in pressure, meanwhile, dynamic moduli depend on the pressure
applied at the wave’s direction of propagation. Based on experimental triaxial test and sonic
27
wave velocity, Yale and Jamieson (1994) indicated that the difference between static and
dynamic moduli lies on mineralogy and porosity of the rock.
As discussed above, elastic moduli play an important role in the mechanical behaviour of
the rock, and these can be static or dynamic depending on the method used for estimating them.
However, static methods are expensive and take up many hours of work in the laboratory. In
contrast, dynamic methods are cheap and easily applied. Despite this, the elastic moduli
estimated through triaxial tests are considered more representative of underground conditions.
Young’s modulus is not only considered one of the most important elastic moduli but
also the elastic modulus that presents the largest difference between static and dynamic
conditions. There are many empirical relations available to convert dynamic moduli into static
moduli. For example Eissa and Kazi (1988), based on stress and strain measurements of 342
laboratory tests with extreme variability on rock types, developed two statistical relations for
estimating Young’s modulus. The first is a linear relation with a correlation coefficient (r2) equal
to 0.84, and the second is a logarithmic correlation with a correlation coefficient equal to 0.96.
Wang (2000) presented a linear correlation applicable to soft and hard rocks. Canady (2011)
presented a non-linear correlation to correct Young’s modulus in formations that vary from soft
to hard rocks. Young’s modulus expression is a modification of Wang’s correlation where
coefficients a, b and c are 1, -2 and 4.5, respectively. Morales and Marcinew (1993) presented a
correlation to correct Young’s modulus taking into account the degree of consolidation, porosity
and mineralogy. Table 2-2 summarizes the most common relationships mentioned above.
28
Values of Est= static and Edy= dynamic are expressed in GPa and the density in g/cm3
Another important mechanical property of the rock that has to be corrected from dynamic
conditions to static conditions is Poisson’s ratio, which is an indicator of the degree of rock’s
consolidation and relates the principle strains of the rock. The theoretical value for a porous rock
saturated with uncompressible fluid is expected to be 0.5. However, according to Spencer et al.
Table 2-2 Corrections to convert dynamic into static moduli.
Rock Type Empirical Correlation Author Wide Range 𝐸𝐸𝑠𝑠𝑠𝑠 = 0.74𝐸𝐸𝑑𝑑𝑦𝑦 − 0.82 Eissa and Kazi
Wide Range 𝐿𝐿𝐿𝐿𝐿𝐿10𝐸𝐸𝑠𝑠𝑠𝑠 = 0.77𝐿𝐿𝐿𝐿𝐿𝐿10�𝜌𝜌 ∗ 𝐸𝐸𝑑𝑑𝑦𝑦� + 0.02 Eissa and Kazi
Soft Rock 𝐸𝐸𝑠𝑠𝑠𝑠 = 0.41𝐸𝐸𝑑𝑑𝑦𝑦 − 1.06 Wang
Wide Range 𝐸𝐸𝑠𝑠𝑠𝑠 =
𝐿𝐿𝐿𝐿�𝐸𝐸𝑑𝑑𝑦𝑦 + 𝑎𝑎� ∗ �𝐸𝐸𝑑𝑑𝑦𝑦 + 𝑏𝑏�𝑐𝑐
Canady
Wide Range 𝐿𝐿𝐿𝐿𝐿𝐿10𝐸𝐸𝑠𝑠𝑠𝑠 = 𝐴𝐴1𝐿𝐿𝐿𝐿𝐿𝐿10�𝐸𝐸𝑑𝑑𝑦𝑦�+ 𝐴𝐴𝐿𝐿 Morales
Berea Sandstone 𝐾𝐾𝑠𝑠𝑠𝑠 = 0.9𝐾𝐾𝑑𝑑𝑦𝑦 Cheng
Wide Range 𝐾𝐾𝑑𝑑𝑦𝑦 =
𝐾𝐾𝑠𝑠𝑠𝑠1 + 3𝑃𝑃𝐾𝐾𝑠𝑠𝑠𝑠
Fjaer
29
(1994) extensive experimental results have indicated that Poisson’s ratios for clean dry sandstone
are in the range of 0.15 to 0.20 when using ultrasonic techniques. Karig (1996) developed static
tests on sediments from the Nankai Channel that resulted in values ranging from 0.20 to 0.25
(cited by Chang and Zoback, 1998).
Chang et al. (1999) indicated that experiments made on the Lentic sand, Wilmington
sand, and Ottawa sand/Montmorillonite samples present viscoelastic behaviour (materials that
present both viscous and elastic properties when undergoing deformation), which cannot be
linked with pore fluid expulsion or dewatering. Therefore, there is a clear difference observed
between static and dynamic moduli for poorly consolidated sand, which can be attributed to a
viscoelastic mechanism.
2.4 Rock Strength
2.4.1 Failure criterion
Failure is the resulting displacement of two sides of a failure plane relative to each other
due to shear stresses large enough to exceed the frictional force that presses the sides together
(Figure 2-3). Then, the critical shear stress (τmax) is a function of normal stresses (σ) acting in
the failure plane. This statement is known as the Mohr’s hypothesis (Fjaer et al. 2008).
Figure 2-3 Failure plane resulting from shear and normal stresses.
σ1
σ3σ
Plane of Shear stress
θ
τ
30
There are three principal stresses acting on an underground unit element of rock, σ1, σ2
and σ3, and the way to connect them is through the Mohr’s circle. Shear failure depends on the
maximum and minimum principal stresses. The normal and shear stresses act on a plane whose
normal makes an angle of θ degrees to the maximum principal stress, σ1 (Figure 2-3).
The Tresca criterion states that rock will yield when the shear stress exceeds the critical
shear stress and the rock will not return to its initial state after removing the stresses. This
behavior is governed by the following expression,
𝜏𝜏𝑚𝑚𝑚𝑚𝑥𝑥 = 0.5 ∗ (𝜎𝜎1 − 𝜎𝜎3) = 𝛿𝛿𝑣𝑣 Eq. 2-30
Where So is the shear strength of the rock also known as cohesion. In Figure 2-4, the Tresca
criterion would be represented as a straight horizontal line.
2.4.2 Coulomb criterion
Coulomb (1785) based on his study of friction force along a sliding plane between two no
welded bodies, and in analogy with the shear stress that cause failure in the rock along a failure
plane, stated that the friction force acting against displacement is equal to the normal force acting
along this plane multiplied by a factor equivalent to tan φ (cited in Jaeger et al., 2007). Also,
shear stress is affected by an additional resistance produced by an internal force of the rock
known as cohesive force (So). Thus, Coulomb criterion is expressed as follows,
𝜏𝜏 = 𝛿𝛿𝑣𝑣 + 𝜎𝜎 tan𝜑𝜑 Eq. 2-31
Where φ is the internal friction angle and is defined as the angle between the failure plane and
horizontal plane.
According to Jaeger et al. (2007), Coulomb’s theory has two limitations. First,
experimental data indicate that σ1 at failure increases at nonlinear rate with σ3. On the contrary,
31
Coulomb’s theory suggests that σ1 at failure will be linear with σ3. Second, Coulomb’s criterion
can be applicable only when σ3=σ2.
The Mohr theory indicates that it is possible to define a semicircle for any state of stress
that results tangent to the straight line that represents the plane where shear and normal stresses
satisfies Coulomb’s criterion. Therefore, Mohr (1914) suggested a more general expression to
represent the failure envelope (cited in Jaeger et al., 2007).
The failure envelope can be experimentally established by plotting a series of Mohr’s
circles for the stresses when the rock fails. The performance of three or more triaxial tests at
different confining pressure allows to build Mohr’s envelope (strength envelope), which is
represented by a straight-line that refers to the limit of strength. As shown in Figure 2-4, the
circle defined by σ1 and σ3 and the envelope has the form τ= f(σ).
Mohr’s strength theory involves two key assumptions:
1) The maximum horizontal stress does not have effect on the strength
2) There is no cohesion, which means that just the pressure has bearing on
the shearing strength.
Figure 2-4 Failure stress.
32
The Mohr-Coulomb criterion is the most used criterion to estimate the strength of the
rock and it is supported by the assumption that strength is a linear function of normal stress as
presented in Eq. 2-32,
𝜏𝜏 = 𝛿𝛿𝑣𝑣 + 𝜎𝜎 tan𝜑𝜑 = 𝛿𝛿𝑣𝑣 + µ Eq. 2-32
Where the coefficient of internal friction (μ) is expressed by μ=tan ϕ and the inherent shear
strength of the rock (So) is measured when the internal friction angle is zero.
The Mohr’s circles shown in Figure 2-5 touch the failure envelop at a specific point of
normal stress and shear stress given by the following expressions, respectively,
𝜎𝜎 =(𝜎𝜎1 − 𝜎𝜎3)
2+
(𝜎𝜎1 − 𝜎𝜎3)2
cos 2𝜃𝜃 Eq. 2-33
𝜏𝜏 =(𝜎𝜎1 − 𝜎𝜎3)
2sin 2𝜃𝜃 Eq. 2-34
Where:
σ = Normal stress acting on the plane of failure.
τ = Shear stress acting on the plane of failure.
2θ = Inclined plane angle.
Once σ and τ are known, it is possible to find the plane of failure where shear stress is
equal to shear strength. Mohr’s circle of stress in Figure 2-5 helps to determine if a failure will
occur or not on a rock along a plane inside the rock and what will be the angle of failure. 2θ is
the angle between the straight line given by the points where the failure envelope touches the
Mohr’s circle and the centre of Mohr’ circle as shown in Figure 2-5. θ provides the direction of
the failure plane and is related to the internal friction angle as follows,
33
θ =𝜋𝜋4
+𝜑𝜑2
Eq. 2-35
Figure 2-5 Mohr’s envelope.
The rock’s strength estimation is fundamental to evaluate common reservoir problems in
the petroleum industry such as compaction, subsidence and sand production.
The rock’s strength is defined as the value of axial compression stress at which the rock
cannot support more stress and as a consequence fails. One of the most important parameters
used to measures the rock strength is the unconfined compressive strength (UCS), which can be
determined in two different ways: directly from both triaxial compression test or uniaxial
compressive test and application of Eq. 2-36, or indirectly from wireline logs and correlations.
The expression for unconfined compressive strength (UCS) is,
𝑈𝑈𝑈𝑈𝛿𝛿 = 2𝛿𝛿𝑣𝑣 tan𝜃𝜃 Eq. 2-36
A uniaxial compressive test is that where the sample is compressed axially without any
radial confining pressure and the UCS value is measured at the point when the rock fails. It
should be noticed that UCS measurements are highly sensitive to cracks and heterogeneities.
Therefore, high experimental uncertainty should be expected (Fjaer et al., 2008; Wood and
Shaw, 2012).
34
In a triaxial compression test the sample is initially compressed axially until the stress
reaches the confining pressure. Then, the confining pressure is held constant and the axial
pressure is increased until the rock fails. It is important to mention that triaxial tests can be
developed under drained and undrained conditions and these allow to measure UCS, Young’s
modulus, Poisson’s ratio, shear modulus and bulk modulus of the framework (Fjaer et al., 2008).
The unconfined compressive strength (UCS) can be determined from triaxial tests on
cylindrical rock samples. However, UCS data are scarce in many reservoirs and much more
under overburden and underburden formations. Thus, empirical relations to estimate the UCS
and internal friction angle (φ) from geophysical well logs have been published by several authors
as alternate solutions. For example, Chang et al. (2005) compiled and summarized some of the
most important empirical relations for calculating UCS and internal friction angle for sandstones
and shales based on internal properties of the rock such as Young’s modulus, porosity, density
and acoustic velocity.
2.4.3 Relations between rock strength and physical properties
The rock strength depends on internal and external factors of the rock. Several authors
have indicated that the most important internal factors governing the UCS for clastic rocks are
composition, clay content, porosity, texture, degree of cementation, discontinuities, water
content, frictional sliding between grains and macroscopic tensile cracks. Meanwhile, the
external factors are confining pressure and temperature (Plumb, 1994; Li et al., 2012; Tziallas et
al., 2013; Brace et al., 1966). Furthermore, in brittle rocks cracks occur abruptly due to the lack
of strength to support compression. On the other hand, in a ductile material the failure process is
gradual (Zoback, 2006).
35
An easy and practical way for estimating UCS is the application of empirical relations as
a function of the aforementioned parameters. Plumb (1994), based on 784 unconfined
compressive tests on sedimentary rocks, observed that UCS increases as porosity decreases and
porosity decreases as Vclay content decreases. Then, he developed a correlation to estimate UCS
as a function of porosity. Vernik et al. (1993) developed similar correlations to Plumb’s equation
but they included an additional parameter ξ to consider pore structure. Parameter ξ is the
reciprocal value of porosity when UCS =0. Wood and Shaw (2012) presented a methodology to
correlate UCS to dry density for sandstones based on experimental data of 134 sandstone tests
and 129 siltstone tests. Tziallas et al., (2013) developed a study on heterogeneous rocks with
hard and weak alternate layers and found that Young’s modulus and siltstone content correlate
well with UCS. They also indicated that sandstones present a USC range between 85 and 103
MPa, whereas siltstone range is between 34.6 and 53.7 MPa. They suggested that the decrease in
strength of a composite specimen is directly related to the percentage of siltstone. Finally, Chang
et al. (2005) proposed a correlation for weak sandstones using wireline sonic logs.
2.5 Coupling Reservoir-Geomechanical Models
In the petroleum industry, reservoir simulation models have been used during many years
to simulate a wide variety of reservoir’s phenomena. However there are some issues of practical
importance, for instance, compaction, subsidence and sand production that cannot be adequately
predicted by only the use of traditional simulation models. The common characteristic of these
issues is the strong relationship between the porous medium, saturation fluids and stress-strain
behaviour of the rock. Therefore, a better way to simulate these problems is required and it is
achieved through the coupling of reservoir models to mechanical models.
36
An important question that reservoir engineers have to answer before initialization of the
coupling of reservoir and geomechanical models is “Is it necessary to consider the effect of
mechanical properties of the rock to reproduce or model these phenomena?” According to
Settari et al. (2000) the answer to this question could be obtained from field and laboratory data.
Settari (2008) stated the following problems where geomechanical models are useful to model
reservoir’s behaviour,
• Soft sands where the strength of the rock is low and the main mechanism of
production is depression. Some indications of these types of formations are low
recovery, high production of sand, and permeability and porosity stress-
dependent (well testing diagnostic).
• Reservoirs presenting compaction and subsidence.
• Reservoirs where porosity and permeability vary with stresses.
• Low gas permeability reservoirs with stress sensitive observation.
• Hydraulic fracturing of conventional and unconventional reservoirs.
• Water injection at pressures equal to or greater than fracture pressure.
• Double porosity reservoirs with low fracture permeability.
• Oil sands with high sand production.
• Chalk reservoirs.
• Completion of wells in coalbed methane formations (CBM).
Additionally, Shchipanov et al. (2010) stated that some important parameters that can
help determining the stress dependency of a reservoir are production history, sample test
37
analyzes, well testing analyses and low accuracy in history matching and production forecasting
processes of reservoir simulation when using only reservoir simulation models.
The concept of coupling numerical modeling (reservoir and geomechanical models) is
complex and involves three different areas: 1) reservoir simulation to predict fluid flow and heat
transfer in porous media, 2) geomechanical simulation to calculate the stress and strain behaviour
of the rock, and 3) failure mechanics, which handles fault geometry and propagation (Settari and
Maurits, 1998).
From the construction’s model point of view, there are three types of coupled problems:
1) regional models, 2) full field reservoir models, and 3) single well models. This work focuses
on full field reservoir simulation. These models aim at the investigation of the following
objectives: determine the oil recovery factor by compaction mechanism, quantify the amount of
subsidence and the size of the area under the effect of subsidence, determine shear stress
distribution to minimize infill drilling, verify fault stability, design hydraulic fracturing jobs
considering changes in stresses, and test the effect of shallow aquifers in final recovery factor.
Settari and Walters (1988), Settari and Sen (2008) and Rodriguez (2011) stated that the
coupling modeling should consider the following stages:
• Definition of boundary conditions and solution domain,
o Reservoir: Constant pressure or closed boundary
o Geomechanics: Do not allow vertical movement at the top of the model, at
the bottom of the model, and at the sides in the normal direction
• Gridding characteristics: Geomechanical grids must consider a side-burden at
least 3 to 5 times the size of the reservoir model, in offshore reservoirs
overburden from the sea floor (but taking into account the effect of sea water), in
38
onshore reservoirs overburden from the ground surface, and underburden as large
as possible. The stress model requires to be populated with geomechanical
properties of the rock. Both geomechanical and reservoir grids should be
independent but compatible.
• Model initialization.
• Selection of coupling method.
According to Tran et al. (2002), there are four methods available in the literature to
couple reservoir flow model to rock stress-strain models. These methods are briefly described as
follows:
1) Explicit coupled (one way coupling method): The governing equations of both
models are solved at each time-step considering the last existing calculations of the
coupling term. One-way indicates that any alteration in pore pressure will cause
alterations in the stress-strain behaviour. However, no alteration will occur in pore
pressure when alteration on stress-strain occurs. The main disadvantage of this
method is thus that it does not consider porosity and effective permeability alterations
as the stresses in the rock change.
Theoretically this method is appropriate when the reservoir compaction is small.
There is not significant error since rock compressibility is governed primarily by gas
compressibility. Consequently, the mass balance is ruled by gas pressure instead of
stresses in the rock. However, this method is not applicable in stress-sensitive gas
reservoirs because dynamic permeability affects fluid flow in porous medium
(Gonzalez, 2012).
39
2) Iteratively Coupled: This two-way coupling scheme satisfies both flow and stress-
strain governing equations of the models. Coupling terms (porosity and permeability)
are iterated during each time step. In this method the data are explicitly passed forth
and back while waiting for achieving convergence tolerance. For example, if volume
coupling is the target, the information passing from the reservoir simulator to the
geomechanical module is pressure and temperature, which are considered as external
loads for displacement calculations. Whereas the information from the geomechanical
module to reservoir simulator is a porosity function, which is computed after stresses
and strains are calculated. The number of iterations is a function of the tolerance
criterion on pressure or stress changes between the last two iterations. An iterative
method is recommended for cases where the compressibility of the rock could
produce a strong deformation on the rock (Tran et al., 2002).
3) Fully coupled: This method implies simultaneous solution of a system of equations
with displacement, temperature and pressure as unknowns. This model is
recommended for modeling complex problems such as plasticity.
4) Decoupled: Commercial reservoir simulators have options to solve fluid flow
problems by themselves since they evaluate stress changes through pore volume
changes and vertical displacements. Porosity and permeability are updated in the
simulator by means of introduction of tables of these variables as a function of
pressure. This method runs fast, easy to use and could be utilized as a first
approximation for preliminary understanding of the reservoir physics.
Consideration of the stress-strain effects in a reservoir simulator is a very difficult and
time consuming task due to the following three key reasons: 1) the lack of reliable input data for
40
the geomechanical modeling, which can lead to misleading results, 2) the huge number of grid
cells required to model the reservoir volume and its surroundings, and 3) the required coupling
method to model the mechanical behavior of the reservoir and surroundings. Hence, it is highly
recommended to analyze the existing direct and indirect input data for determining if the
reservoir stress-strain dependency is significant (Gonzalez, 2012).
2.6 Reservoir Compaction and Subsidence
Compaction is an irreversible mechanism that produces the rock’s shrink by means of
increasing the effective stress when there is a reduction in reservoir pore pressure. As a result,
the stresses exceed the compressive strength of the rock and consequently porosity and
permeability are permanently reduced. The important consequence above the reservoir is the
seabed or ground surface deformation also known as subsidence.
Morita (1989) indicated that in-situ stresses change as pore pressure changes; then, the
rock deforms axially with horizontal strain. He presented a methodology that allows estimating
the degree of compaction and subsidence in reservoirs with a simple geometry by assuming that
in-situ stresses are composed of induced and original in-situ stresses. The methodology provides
a good approximation but it is limited to reservoirs with low Young’s modulus.
Settari in 2002 indicated that compaction is subjected to the initial stress state on the rock
and the stress path developed during reduction of pressure once the reservoir goes on production.
When the mechanism of compaction is only a function of the mean effective stress, the decrease
in porosity follows a smooth trend during the depletion process. However, if the compaction
process involves shear failure as in the case of chalk reservoirs or when temperature increases,
then porosity is a function of a combination of stresses and more complex stress-strain models
should be used to simulate the volumetric strain response. In Figure 2-6a the elastic behavior of
41
the rock can be approached based on the rock compressibility. Plastic deformation starts once the
compressive strength exceeds point A. Following plastic deformation to point B any unload at
this point will follow a hysteresis path and the slope will be less than for the initial elastic
loading. In Figure 2-6b the same process is presented using a cap model. Note that the pore
collapse mechanism occurs when the stress path reaches the cap envelope.
Figure 2-6 (a) Rock compaction versus effective stress, (b) Rock compaction in Mohr-Coulomb diagram. (Source: Settari, 2002)
42
Mechanical Properties and In-Situ Stresses
This chapter describes the methodology followed to estimate the in situ stresses and pore
pressure at reservoir conditions and to evaluate the situ rock mechanical properties using sources
of information such as core samples, wireline logs, and well tests. New empirical correlations are
developed in this thesis to convert dynamic moduli to static moduli for the case of Young’s
modulus, Poisson’s Ratio, unconfined compressive strength and friction angle of rocks in the
area of study.
3.1 Static and Dynamic Moduli
As previously mentioned in Chapter Two, there are two key sources of information for
determining rock mechanical properties: 1) direct measurements from triaxial tests in the
laboratory that provide properties known as “static moduli”, and 2) calculations using wireline
logs such as bulk density, compressional acoustic velocity and shear acoustic velocity that
provide properties known as “dynamic moduli”. The static elastic moduli are different from the
dynamic elastic moduli since they are determined under different conditions; however, they can
be related. The first step to relate these two sources of information is to correct the static moduli
on the basis of effective in-situ stress and then correlate it to the dynamic moduli. In order to
correct the results of triaxial test at reservoir conditions, the confining stresses must be
determined at reservoir conditions.
3.2 Experimental Data
Triaxial tests on samples from wells GML-2 (located in the field of interest), GMK-1 and
GMK-2 (located in a neighbouring field) were performed using a triaxial cell equipped with
gauges that control the radial confining pressure, screw-ball press for high precision on axial
load, strain-gauges in the load cell, linear variable differential transformer (LVDTs) to measure
43
radial deformation, ultrasonic sources to generate P and S waves, and detectors to measure the
porosity, grain density and waves velocities in different directions. Figure 3-1 presents an
illustration of a triaxial cell (not to scale) similar to that used for GML elastic moduli
determination.
Figure 3-1 Triaxial cell (Source: FHWA, 2007) The testing procedure considered four stages:
1) Collection of 3 or 4 plugs at every depth of interest to guarantee the Mohr-
Coulomb envelope’s reliability.
44
2) Preparation of cylindrical plugs (1 inch in diameter and 2 inches in length) and
placing of them in the triaxial cell.
3) Scanning of samples with a CT-scan to guarantee their homogeneity and integrity.
4) Testing of samples under different confining pressures. At this point the samples
are hydrostatically pressurized until reaching the confining pressure. Measure
acoustic velocities (acoustic tests were only carried out for wells GMK-1 and
GMK-2).
5) Holding the confining pressure constant increase the axial load until the rock fails.
Mechanical and acoustic information were gathered from these tests. Some of the
mechanical information includes confining pressure, axial stress, axial strain and lateral strain.
The acoustic measurements include P and S wave velocities.
The case of a poroelastic rock with a nonlinear stress-strain relationship is shown in
Figure 3-2a. For this case, the static bulk modulus can be estimated from the hydrostatic phase
since the stresses are equal in all directions and the relation between stress and strain is linear.
The Young’s modulus can be determined using the tangent method, where the slope of a straight-
line tangent to the stress-strain curve in the triaxial phase is equal to Young’s modulus. Poisson’s
ratio is equal to the ratio between the radial stress-strain slope and the axial stress-strain in the
triaxial phase. Shear modulus can be calculated from any of the other two moduli determined
previously and the use of one of the expressions listed in Table 2-1.
Figure 3-2a shows the relationship between strain and axial stress for three triaxial tests
under different confining pressures 200, 400 and 800 psi (1.37 MPa, 27.5 MPa 55.15 MPa) and
for well GML-2. However none of them represent the reservoir conditions as will be shown later
in this Chapter. The stress-strain relationship for each test exhibits linear behavior previous to
45
rock failure, which indicates that the rock is acting as an elastic material. It can also be observed
that as long as the confining pressure decreases, the rock behaves as a poroelastic material and as
a consequence the failure stress also decreases.
Figure 3-2b shows Mohr’s circles for three different confining pressures and the Mohr-
Coulomb failure line. The failure line allows estimating the inherent shear strength (cohesion)
and the internal friction angle.
Figure 3-2 Triaxial test results. (a) Stress versus strain and (b) Mohr-Coulomb circle with failure line.
Similar plots were built for each one of the 30 sets of triaxial tests. Table 3-1, Table 3-2
and Table 3-3 show the calculated values of bulk modulus, shear modulus, Poisson’s ratio and
Young’s modulus for both static and dynamic conditions. These tables also present information
for the triaxial tests including well name, sampling’s depth, facies, porosity and confining
pressure. Sonic velocity test data from laboratory is also included for wells in GMK field.
0200400600800
100012001400160018002000
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Shea
r Str
ess (
Psi)
Normal Stress (Psi)
Mohr-Coulomb Circles
0
500
1000
1500
2000
2500
3000
3500
0 0.5 1 1.5 2 2.5 3
Stre
ss (p
si)
Strain (%)
Stress vs. Strain
Confining pressure 200 psi
Confining pressure 400 Psi
Confining pressure 800 Psi
Confining Pressure 1600 Psi
(a) (b)
Friction Angle =22.8Cohesion = 454.2 psi
46
Table 3-1 GML experimental data.
Depth Confining Axial PoissonPressure Strengh Young Shear Bulk Ratio
m Psi Psi 106 Psi 106 Psi 106 Psi
GML-2 N2MC14A Shale 3057.43 200 2270 223.0 88.3 153.2 0.262GML-2 N2MC14C Shale 3057.43 400 2882 249.5 102.8 145.1 0.213GML-2 N2MC14B Shale 3057.43 800 4144 558.0 235.3 295.9 0.186GML-2 N3MC1G Sand 3072.98 200 1373 1790.0 695.0 1406.0 0.288GML-2 N3MC1H Sand 3072.98 400 2038 2320.0 916.5 1650.0 0.266GML-2 N3MC1I Sand 3072.98 800 2600 2930.0 1170.0 1970.0 0.252GML-2 N3MC5E Sand 3078.54 200 1355 185.0 74.4 120.1 0.243GML-2 N3MC5F Sand 3078.54 400 2059 262.7 109.4 146.2 0.200GML-2 N3MC5L Sand 3078.41 800 2615 371.2 156.3 197.9 0.187GML-2 N3MC5N Sand 3078.41 1600 3215 453.3 191.9 236.6 0.181GML-2 N3MC5C Sand 3078.54 200 1393 160.0 61.9 128.3 0.292GML-2 N3MC5D Sand 3078.54 400 1779 244.8 101.6 137.9 0.204GML-2 N3MC5O Sand 3078.41 1600 3250 411.9 178.6 198.0 0.153GML-2 N4MC1A Sand 3083.60 200 5275 1100 465.3 576.5 0.182GML-2 N4MC1C Sand 3083.65 400 8630 1450 614.9 752.9 0.179GML-2 N4MC1D Sand 3083.65 800 12048 1900 814.1 951.0 0.167GML-2 N4R5H2 Sand 3086.52 200 1522 183.0 72.0 129.0 0.263GML-2 N4R5H3 Sand 3086.56 400 1741 188.0 76.0 116.0 0.230GML-2 N4R5H4 Sand 3086.61 800 2433 308.0 129.0 166.0 0.191GML-2 N4R5H5 Sand 3086.65 1600 3809 312.0 136.0 149.0 0.151GML-2 N4MC3A Sand 3087.83 200 1702 253.0 98.8 191.7 0.280GML-2 N4MC3B Sand 3087.83 400 2319 315.0 123.7 231.3 0.273GML-2 N4MC3C Sand 3087.83 800 2951 456.0 184.5 287.9 0.236GML-2 N5MC1H Shale 3175.17 200 1978 204.0 82.6 128.3 0.235GML-2 N5MC1I Shale 3175.17 400 2358 215.0 87.4 132.8 0.230GML-2 N5MC1K Shale 3175.17 800 3201 314.0 137.8 145.0 0.139
Poroelastic moduliWell
Rock TypePlug
47
Table 3-2 GMK-1 experimental data.
Confining Axial Poisson PoissonDepth Pressure Strengh Young Shear Bulk Ratio Compres. Shear Young Shear Bulk Ratio
m Psi Psi 106 Psi 106 Psi 106 Psi ft/sec ft/sec 106 Psi 106 Psi 106 Psi
GMK-1 N1DC8V12 Shale 0.211 2856.7 100 684 0.055 0.021 0.049 0.313 3019 1807 1.67 0.237 0.097 0.142 0.220GMK-1 N1DC8V2 Shale 0.190 2856.6 500 872 0.068 0.028 0.040 0.218 3027 1828 1.78 0.243 0.095 0.175 0.283GMK-1 N1DC8V4 Shale 0.240 2856.6 900 927 0.044 0.018 0.024 0.203 3056 1845 1.66 0.248 0.102 0.144 0.212GMK-1 N1DC8V1 Shale 0.180 2856.6 1300 1119 0.036 0.016 0.018 0.158 3100 1905 1.63 0.259 0.108 0.142 0.196GMK-1 N2DC18H1 Shale 0.011 2885.15 100 464 0.036 0.013 0.038 0.346 2452 1468 1.67 0.144 0.059 0.086 0.220GMK-1 N2DC18H3 Shale 0.010 2885.18 500 661 0.020 0.007 0.019 0.331 2906 1674 1.74 0.199 0.080 0.134 0.251GMK-1 N2DC18H2 Shale 0.009 2885.15 900 904 0.018 0.007 0.012 0.255 2927 1750 1.67 0.210 0.086 0.126 0.222GMK-1 N2DC18H4 Shale 0.006 2885.21 1300 1029 0.018 0.007 0.011 0.230 2933 1753 1.67 0.217 0.089 0.130 0.222GMK-1 N4CV85C Toba 0.255 3951.62 100 6037 0.886 0.409 0.354 0.083 9944 5425 1.83 1.997 0.775 1.571 0.288GMK-1 N4CV85A Toba 0.246 3951.62 600 6534 0.941 0.437 0.370 0.076 9735 5344 1.82 1.987 0.774 1.535 0.284GMK-1 N4CV85B Toba 0.254 3951.62 1100 6672 0.953 0.434 0.396 0.099 9835 5418 1.82 1.996 0.778 1.527 0.282GMK-1 N4CV85D Toba 0.254 3951.62 1600 6663 1.004 0.470 0.386 0.067 9947 5462 1.82 2.027 0.789 1.565 0.284GMK-1 N4V4B Sand 0.102 3916.44 100 8385 1.328 0.529 0.905 0.255 14043 7771 1.81 5.044 1.972 3.808 0.279GMK-1 N4V4A Sand 0.100 3916.44 600 10479 1.739 0.710 1.051 0.224 14383 8173 1.76 5.529 2.192 3.865 0.262GMK-1 N4V4C Sand 0.102 3916.44 1100 12596 1.909 0.804 1.016 0.187 14664 8495 1.73 5.927 2.376 3.911 0.247GMK-1 N4V4D Sand 0.102 3916.44 1600 14643 2.333 0.987 1.223 0.182 14644 8768 1.67 6.171 2.528 3.679 0.220GMK-1 N4V34B Sand 0.192 3920.37 100 1423 0.210 0.075 0.379 0.408 8965 4323 2.07 1.525 0.565 1.678 0.349GMK-1 N4V34C Sand 0.188 3920.37 600 3039 0.334 0.134 0.216 0.243 9726 4778 2.04 1.892 0.705 1.982 0.341GMK-1 N4V34D Sand 0.191 3920.37 1100 3462 0.355 0.142 0.239 0.252 9850 4845 2.03 1.947 0.726 2.033 0.340GMK-1 N4V34E Sand 0.193 3920.37 1600 4093 0.441 0.178 0.280 0.237 10047 5166 1.94 2.182 0.826 2.024 0.320GMK-1 N4CV12A Sand 0.152 3936.2 100 2268 0.340 0.127 0.350 0.338 8959 3951 2.27 1.358 0.492 1.876 0.379GMK-1 N4CV12B Sand 0.150 3936.2 600 3370 0.498 0.199 0.329 0.248 9828 5102 1.93 2.182 0.829 1.971 0.316GMK-1 N4CV12C Sand 0.150 3936.2 1100 4090 0.810 0.324 0.540 0.250 9992 5359 1.86 2.379 0.916 1.964 0.298GMK-1 N4CV12F Sand 0.153 3936.2 1600 4447 0.622 0.252 0.388 0.233 10017 5402 1.85 2.420 0.934 1.967 0.295GMK-1 N4CV30A Sand 0.204 3939.58 100 2466 0.329 0.124 0.312 0.324 8481 4842 1.75 1.809 0.719 1.247 0.258GMK-1 N4CV30C Sand 0.203 3939.58 600 3562 0.455 0.177 0.358 0.288 9065 5320 1.70 2.160 0.873 1.371 0.237GMK-1 N4CV30D Sand 0.203 3939.58 1100 4126 0.541 0.221 0.324 0.222 9481 5637 1.68 2.415 0.984 1.472 0.227GMK-1 N4CV30E Sand 0.205 3939.58 1600 4836 0.674 0.278 0.393 0.214 9723 5817 1.67 2.560 1.048 1.531 0.221GMK-1 N5V22A Sand 0.233 4011.8 100 1239 0.123 0.049 0.081 0.248 8842 4764 1.86 1.805 0.697 1.471 0.295GMK-1 N5V22B Sand 0.229 4011.8 600 2577 0.294 0.121 0.174 0.218 9382 5017 1.87 2.038 0.784 1.696 0.300GMK-1 N5V22C Sand 0.233 4011.8 1100 3138 0.329 0.136 0.188 0.208 9747 5364 1.82 2.285 0.891 1.753 0.283GMK-1 N5V22D Sand 0.231 4011.8 1600 3839 0.353 0.149 0.186 0.183 9946 5558 1.79 2.466 0.969 1.810 0.273GMK-1 N5V53A Sand 0.173 4019.2 100 1645 0.166 0.065 0.127 0.282 9970 4684 2.13 1.910 0.703 2.248 0.358GMK-1 N5V53C Sand 0.169 4019.2 600 3271 0.372 0.150 0.240 0.241 10579 4846 2.18 2.082 0.761 2.613 0.367GMK-1 N5V53E Sand 0.171 4019.2 1100 4211 0.431 0.178 0.247 0.209 10863 5345 2.03 2.508 0.935 2.616 0.340GMK-1 N5V53F Sand 0.172 4019.2 1600 5037 0.531 0.222 0.292 0.197 10938 6266 1.75 3.233 1.287 2.206 0.256GMK-1 N5V59A Sand 0.241 4019.89 100 1033 0.094 0.039 0.056 0.220 8680 3817 2.27 1.177 0.426 1.637 0.380GMK-1 N5V59B Sand 0.240 4019.89 600 2576 0.185 0.077 0.102 0.198 9930 4767 2.08 1.911 0.707 2.127 0.350GMK-1 N5V59C Sand 0.239 4019.89 1100 3610 0.235 0.100 0.122 0.179 10218 5038 2.03 2.128 0.794 2.209 0.339GMK-1 N5V59D Sand 0.240 4019.89 1600 4681 0.271 0.115 0.139 0.175 10310 5656 1.82 2.586 1.006 2.002 0.285GMK-1 N5V78A Sand 0.253 4022.18 100 1390 0.159 0.063 0.112 0.263 8514 4210 2.02 1.438 0.537 1.481 0.338GMK-1 N5V78B Sand 0.249 4022.18 600 2574 0.273 0.110 0.173 0.237 9036 4712 1.92 1.792 0.682 1.599 0.313GMK-1 N5V78C Sand 0.248 4022.18 1100 3444 0.341 0.141 0.196 0.210 9224 4835 1.91 1.892 0.722 1.665 0.311GMK-1 N5V78D Sand 0.251 4022.18 1600 3877 0.433 0.181 0.236 0.194 9369 4946 1.89 1.971 0.754 1.701 0.307
STATIC PROPERTIES DYNAMIC PROPERTIES
Well Plug Facie Porosity
Poroelastic moduli Wave Velocity
Relación Vp/Vs
Poroelastic moduli
48
Table 3-3 GMK-2 experimental data.
3.3 Mean Effective Stress
In order to establish a relationship between static and dynamic moduli, the first step is to
define the mean effective stresses in the reservoir. According to Zoback (2006), three principal
stresses have to be known to completely represent the state of stress of a reservoir: the principal
vertical stress perpendicular to the earth surface represented by the overburden weight (Sv), and
two main stresses in the horizontal direction expressed as maximum horizontal stress (SHmax) and
minimum horizontal stress (Shmin). The principal stresses magnitude indicates the faulting
regimes. In normal faulting Sv is the maximum stress, in a strike-slip regime Sv is the
intermediate stress and in reverse faulting Sv is the minimum stress. For the offshore field under
study, these parameters are calculated for the GML-2 area since this is the only well that has
triaxial tests in the field.
Depth Confining Axial Poisson PoissonPressure Strengh Young Shear Bulk Ratio Compres. Shear Young Shear Bulk Ratio
m Psi Psi 106 Psi 106 Psi 106 Psi ft/sec ft/sec 106 Psi 106 Psi 106 Psi
GMK-2 N1R3V3A Sand 0.168 4261.93 150 1697 0.158 0.064 0.098 0.231GMK-2 N1R3V3D Sand 0.171 4261.93 700 3244 0.258 0.104 0.163 0.236GMK-2 N1R3V3C Sand 0.169 4261.93 1250 4272 0.622 0.255 0.373 0.222GMK-2 N1R3V3B Sand 0.165 4261.93 1800 4733 0.442 0.191 0.213 0.154GMK-2 N1R2V1B Sand 0.209 4253.18 150 1159 0.154 0.067 0.075 0.156GMK-2 N1R2V1C Sand 0.208 4253.18 700 2773 0.181 0.08 0.081 0.127GMK-2 N1R2V1A Sand 0.211 4253.18 1250 3985 0.196 0.09 0.08 0.092GMK-2 N1R2V1D Sand 0.2 4253.18 1800 5443 0.270 0.124 0.109 0.087GMK-2 N2R1V5A Sand 0.261 4304.03 150.00 1731 0.213 0.084 0.153 0.268 9056 3818 2.37 1.856 1.204 0.432 0.392GMK-2 N2R1V5B Sand 0.281 4304.03 700 2994 0.338 0.142 0.184 0.193 9741 4326 2.25 2.079 1.536 0.558 0.377GMK-2 N2R1V5C Sand 0.257 4304.03 1250 3712 0.444 0.192 0.214 0.155 9712 4294 2.26 2.073 1.512 0.548 0.378GMK-2 N2R1V5D Sand 0.262 4304.03 1800 4225 0.468 0.204 0.222 0.148 9853 4430 2.22 2.112 1.598 0.582 0.374GMK-2 N2R1V14D Sand 0.182 4310.82 150 1456 0.09 0.04 0.041 0.133 9436 4124 2.29 2.131 1.509 0.546 0.382GMK-2 N2R1V11A Sand 0.204 4310.7 700 3049 0.263 0.118 0.115 0.117 10867 5143 2.11 2.729 2.363 0.871 0.356GMK-2 N2R1V14C Sand 0.181 4310.82 1250 4615 0.236 0.108 0.097 0.096 10090 5052 2.00 2.277 2.285 0.857 0.333GMK-2 N2R1V11B Sand 0.209 4310.7 1800 5095 0.406 0.183 0.175 0.112 10341 5355 1.93 2.286 2.513 0.954 0.317GMK-2 N2R1V20D Sand 0.24 4314.3 150 1469 0.097 0.044 0.041 0.104 9838 5024 1.96 2.000 2.119 0.801 0.324GMK-2 N2R1V20B Sand 0.24 4314.3 700 2986 0.212 0.097 0.087 0.097 10391 5387 1.93 2.141 2.361 0.897 0.316GMK-2 N2R1V20A Sand 0.241 4314.3 1250 4092 0.276 0.126 0.114 0.097 10867 6059 1.79 2.131 2.884 1.132 0.274GMK-2 N2R1V32A Sand 0.259 4318.18 150 1566 0.148 0.065 0.078 0.185 8881 4065 2.18 1.774 1.411 0.516 0.367GMK-2 N2R1V32C Sand 0.26 4318.18 700 2892 0.277 0.121 0.129 0.142 9720 4906 1.98 1.888 1.936 0.728 0.329GMK-2 N2R1V32D Sand 0.259 4318.18 1250 3751 0.341 0.153 0.148 0.115 10196 5119 1.99 2.093 2.116 0.795 0.331GMK-2 N2R1V32E Sand 0.262 4318.18 1800 4595 0.384 0.172 0.168 0.118 10412 5391 1.93 2.125 2.336 0.887 0.317GMK-2 N3R1V2B Sand 0.173 4381.55 100 2027 0.208 0.083 0.143 0.257 8639 4677 1.85 1.449 1.801 0.697 0.293GMK-2 N3R1V2A Sand 0.171 4381.55 800 4776 0.47 0.193 0.279 0.219 9951 5372 1.85 1.942 2.396 0.926 0.294GMK-2 N3R1V5A Sand 0.171 4381.65 1500 6223 0.581 0.238 0.346 0.22 10488 5747 1.83 2.116 2.723 1.060 0.285GMK-2 N3R1V5B Sand 0.175 4381.65 2200 7518 0.69 0.297 0.339 0.161 10753 5904 1.82 2.187 2.832 1.103 0.284GMK-2 N3R1V14A Sand 0.148 4393.62 100 3750 0.458 0.185 0.29 0.237 9093 4616 1.97 1.808 1.887 0.712 0.326GMK-2 N3R1V14B Sand 0.149 4393.62 800 5660 0.735 0.308 0.4 0.194 10285 5617 1.83 2.126 2.710 1.052 0.287GMK-2 N3R1V14C Sand 0.152 4393.62 1500 7516 0.921 0.385 0.502 0.194 10884 6141 1.77 2.275 3.192 1.263 0.265GMK-2 N3R1V14D Sand 0.148 4393.62 2200 9090 1.06 0.446 0.567 0.188 11330 6274 1.81 2.536 3.362 1.314 0.279
DYNAMIC PROPERTIESSTATIC PROPERTIES
Well Plug Rock Type
Porosity Relación Vp/Vs
Poroelastic moduli Poroelastic moduliWave Velocity
49
3.3.1 Pore pressure
Pore pressure was measured using a formation dynamic tester tool and performing
pressure transient analysis for well GML-2. The interpretation indicated a pore pressure equal to
365.6 kg/cm2 at depth of 3078 m (upper reservoir datum).
3.3.2 Vertical stress
Vertical stress is a function of depth and rock density. For the case of onshore reservoirs
it is calculated by integration of rock density from the surface to the upper reservoir datum. In
the case of offshore reservoirs the rock density must be corrected for seabed depth. The
following expression is used for calculating Sv values,
𝛿𝛿𝑣𝑣 = 𝜌𝜌𝑤𝑤𝐿𝐿𝑍𝑍𝑤𝑤 + �̅�𝜌𝐿𝐿(𝑍𝑍 − 𝑍𝑍𝑤𝑤) Eq. 3-1
Where ρw is water density, g is the gravitational acceleration, Z is the depth of interest, Zw is the
water depth, and �̅�𝜌 is the mean overburden density.
At shallow depths where density log data were not available, a pseudo density log was
generated as an exponential function of depth in such a way that it follows the density log trend
at other depths as shown in Figure 3-3. For this case the vertical stress is calculated with the use
of Eq. 3-1 to be 545.14 kg/cm2 at 3078 m as shown in the calculation below.
𝛿𝛿𝑣𝑣 =1020𝑘𝑘𝐿𝐿𝑚𝑚3 ∗
9.81𝑚𝑚𝑠𝑠2
∗ 1200𝑚𝑚 +2250𝑘𝑘𝐿𝐿𝑚𝑚3 ∗
9.81𝑚𝑚𝑠𝑠2
∗ (3078𝑚𝑚 − 1200𝑚𝑚)
Sv = 53.45 MPa=545.14 kg/cm2
50
Figure 3-3 GML-2 bulk density log. 3.3.3 Minimum horizontal stress
The most reliable and accepted method for estimating the minimum horizontal stress
(Shmin) from field data is based on fracturing the rock and recording the closure pressure (leak off
test, LOT). This method requires a radius of penetration into the formation 2 to 3 times larger
than the wellbore radius in order to capture only the internal stress. The LOT is usually carried
out during the drilling stage, after the casing is in place by drilling some additional meters into
the formation. This test helps to determine the fluid density required for the next drilling stage
and is the more common method used for determining the minimum horizontal stress (Fjaer et
al., 2008).
It is important to highlight that hydraulic fracturing can only be used to determine the
magnitude of Shmin in normal faulting or strike-slip environments. GML reservoirs satisfy this
condition as will be demonstrated later in this chapter.
1100
1300
1500
1700
1900
2100
2300
2500
2700
2900
3100
33001 1.5 2 2.5 3
Dep
th (m
)
Bulk density (g/cm3)
51
The minimum horizontal stress of the lower Miocene reservoirs in GML-2 area are
determined from the analysis of 2 leak off tests. The first test was performed at a depth of 2273
m in the lower Pliocene formation and the closure pressure was estimated in 304.5 kg/cm2 (29.86
MPa). The second test was an extended leak off test and it was performed at a depth of 2742 m.
The closure pressure was calculated as 386.3 kg/cm2 (37.88 MPa).These two minimum
horizontal stress values were used to estimate a gradient of 0.17 kg/cm2/m (16.6 KPa/m) as
shown in Figure 3-4. This gradient was used to calculate the minimum horizontal stress at 3078
m resulting in Shmin equal to 461.6 kg/cm2 (45.26 MPa).
Figure 3-4 Minimum horizontal stress profile (green line) and leak off test data (red dots).
3.3.4 Maximum horizontal stress
The stress polygon presented in Figure 3-5 helps to estimate a possible range of
maximum (SHmax) and minimum (Shmin) horizontal stress values corresponding to a pore pressure
profile (Zoback, 2006). Some important considerations to take into account while building this
polygon are presented next. If SHmax ≥ Shmin any possible stress is above the 45° straight line.
1000
1500
2000
2500
3000
35000 200 400 600 800 1000
Dep
th (m
)
Shmin(kg/cm2)
Minimum horizontal stress
Shmin 1D
Leak off test
52
Vertical and horizontal lines intersect at a common point where Sv = SHmax = Shmin. Stress
boundaries are associated with reverse faulting (RF), normal faulting (NF) and strike-slip (SS).
The lowest minimum horizontal stress value in Figure 3-5 is represented by the vertical line
located to the left of the 45° straight line. This vertical line also limits the normal faulting as
predicted by Eq. 3-2. On the other hand, the highest maximum horizontal stress value is
represented by the horizontal line located at the top of Figure 3-5. This horizontal line also limits
the reverse faulting predicted by Eq. 3-3. The diagonal line limiting the strike-slip zone
corresponds to a value of SHmax at which strike-slip faulting occurs for a given value of Shmin and
the red dashed lines represent the maximum horizontal stress as function of breakout angle
(Zoback, 2002).
Figure 3-5 Stress polygon.
𝑁𝑁𝐹𝐹: 𝜎𝜎1𝜎𝜎3
=𝛿𝛿𝑣𝑣 − 𝑃𝑃𝑝𝑝
𝛿𝛿ℎ𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑃𝑃𝑝𝑝≤ �(𝜇𝜇2 + 1)
12 + 𝜇𝜇�
2 Eq. 3-2
𝑅𝑅𝐹𝐹: 𝜎𝜎1𝜎𝜎3
=𝛿𝛿𝐻𝐻𝑚𝑚𝑚𝑚𝑥𝑥 − 𝑃𝑃𝑝𝑝𝛿𝛿𝑣𝑣 − 𝑃𝑃𝑝𝑝
≤ �(𝜇𝜇2 + 1)12 + 𝜇𝜇�
2 Eq. 3-3
SS
RF
NF
Shmin
SHm
ax
Sv
53
𝛿𝛿𝛿𝛿: 𝜎𝜎1𝜎𝜎3
=𝛿𝛿𝐻𝐻𝑚𝑚𝑚𝑚𝑥𝑥 − 𝑃𝑃𝑝𝑝𝛿𝛿ℎ𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑃𝑃𝑝𝑝
≤ �(𝜇𝜇2 + 1)12 + 𝜇𝜇�
2 Eq. 3-4
In addition, Barton et al. (1988) suggested a procedure for calculating SHmax, which
requires the rock strength and wellbore breakout width angle (Wbo) as input data. Eq. 3-5 shows
the expression to estimate SHmax determined by Barton et al. (1988),
𝛿𝛿𝐻𝐻𝑚𝑚𝑚𝑚𝑥𝑥 =�𝑈𝑈𝑈𝑈𝛿𝛿 + 2𝑃𝑃𝑝𝑝 + �𝛥𝛥𝑃𝑃𝑝𝑝�� − 𝛿𝛿ℎ𝑚𝑚𝑚𝑚𝑚𝑚(1 + 2𝑐𝑐𝐿𝐿𝑠𝑠2𝜃𝜃𝑏𝑏)
1 − 2𝑐𝑐𝐿𝐿𝑠𝑠2𝜃𝜃𝑏𝑏 Eq. 3-5
Where:
2θb = π - Wbo
UCS = Unconfined Compression Strength
Pp = Pore pressure
ΔP = Difference between mud pressure and pore pressure
Wbo = Angle of breakout width.
θb = Angle of breakout initiation in reference with SHmax
The application of Eq. 3-5 to estimate SHmax for a given Wbo and UCS results in a line
that represents the maximum horizontal stress required to cause a breakout under the specific
conditions of rock strength and Wbo of interest.
Applying the previous methodology to GML-2 data, a stress polygon is created as
follows,
1) Sv = SHmax = Shmin = 545.14 kg/cm2 @ 3078 m on the 45° straight line.
2) Calculation of minimum horizontal stress assuming normal faulting: The triaxial
test indicated that UCS = 1400 psi = 98.4 kg/cm2 and frictional angle (ϕ)=30.5°
54
for sand. UCS = 1600 psi = 112.5 kg/cm2 and frictional angle (ϕ)=30.4° for shale.
μ= tan (ϕ ). Using Eq. 3-2, Shmin is calculated as follows,
545 − 365.6𝛿𝛿ℎ𝑚𝑚𝑚𝑚𝑚𝑚 − 365.6
= �(0.5772 + 1)12 + 0.577�
2
Shmin=425.4 kg/cm2 (41.7 MPa)
3) Calculation of maximum horizontal stress (SHmax) assuming reverse faulting and
using Eq. 3-3:
𝛿𝛿𝐻𝐻𝑚𝑚𝑚𝑚𝑥𝑥 − 365.6545 − 365.6
= �(0.5772 + 1)12 + 0.577�
2
SHmax= 903.8 kg/cm2 (88.6 MPa).
4) Lines representing the SHmax required creating a breakout in the rock for different
unconfined compressive strength values can be generated using equation Eq. 3-5.
The plot shown in Figure 3-6 represents the stress polygon built with the stresses
previously calculated and the red diagonal line corresponds to the values of SHmax required to
induce breakouts with constant Wbo of 30° in the rock for specific rock strength and different
minimum horizontal stresses.
55
Figure 3-6 GML-2 stress polygon. Based on Figure 3-6, the stress regime for GML field is located in the transition between
normal and strike-slip faulting and the maximum horizontal stress with constant Wbo of 30° and
unconfined compressional strength of 105.5 kg/cm2 (10.34 MPa) is equal to 532.5
kg/cm2 (52.2MPa). The results indicate that the degree of anisotropy is not very significant as the
maximum horizontal stress is 532.5 kg/cm2 (52.2MPa) and the minimum is 461.6 kg/cm2 (45.26
MPa); a ratio of only 1.15.
3.3.5 Mean effective stress
Reservoirs under normal faulting regime are characterized by a maximum principal
vertical stress (Sv), the intermediate stress is the maximum horizontal stress (SHmax) and the least
stress is the minimum horizontal stress (Shmin). So for this case Sv > SHmax > Shmin.
The poroelastic moduli obtained from laboratory tests are determined with the use of
stress-strain relationships, which exhibit non-linear behavior when the porous rock is saturated.
Under this condition, the poroelastic moduli are calculated using a Biot’s coefficient (α) lower
than 1. However, if triaxial tests are performed letting the outlet opened to the atmosphere, then,
300
400
500
600
700
800
900
1000
300 500 700 900 1100
S Hm
ax(k
g/cm
2 )
Shmin (kg/cm2)
56
pore pressure is equal to zero, and Biot’s coefficient is considered equal to 1. Consequently the
poroelastic moduli only depends on Terzaghi effective stress equation expressed as the
difference between the confining pressure (Pc) and pore pressure (Pp) (Detournay & Cheng,
1993; Fjaer et al., 2008). Therefore, assuming α=1, the mean effective stress is expressed as
shown in Eq. 3-6,
𝜎𝜎𝑒𝑒𝑓𝑓𝑓𝑓 =𝛿𝛿ℎ𝑚𝑚𝑚𝑚𝑚𝑚 + 𝛿𝛿𝐻𝐻𝑚𝑚𝑚𝑚𝑥𝑥 + 𝛿𝛿𝑣𝑣
3− 𝑃𝑃𝑝𝑝 Eq. 3-6
Upon substitution of the previously calculated vertical stress, minimum horizontal stress
and maximum horizontal stress, GML mean effective stress at reservoir conditions is,
𝜎𝜎𝑒𝑒𝑓𝑓𝑓𝑓 =461.6 + 532.5 + 545
3− 365.6 = 147.4
kg𝑐𝑐𝑚𝑚2 = 14.45MPa = 2097 psi
3.4 Static Moduli
The maximum confining pressure applied to GML-2 samples during the triaxial tests was
1600 psi. However, the mean effective in-situ stress calculated above is 2097 psi. However, the
comparison between dynamic elastic moduli calculated from acoustic wave velocity data and the
static elastic moduli estimated from triaxial tests is only valid under the same confining pressure
conditions. For this reason, the experimental static elastic moduli are extrapolated to a confining
pressure equal to the reservoir’s mean effective stress.
Young’s modulus is extrapolated using a logarithmic function as follows,
𝐸𝐸 = 𝑎𝑎 ∗ 𝐿𝐿𝐿𝐿(𝑃𝑃𝑃𝑃) − 𝑏𝑏 Eq. 3-7
Where coefficients a and b are functions of the input data that correspond to each set of triaxial
tests. The coefficients of determination (R2) of each equation applied to each well analyzed in
this study vary from 0.84 to 1. Figure 3-7a presents the extrapolated Young’s modulus
corresponding to 2100 psi associated with well GML-2.
57
Figure 3-7 Elastic moduli. (a)Young’s modulus extrapolation and (b) Poisson’s ratio extrapolation.
Poisson’s ratio extrapolation is done by means of a power function with the general
expression indicated in Eq. 3-8.
𝑣𝑣 = 𝑎𝑎 ∗ 𝑃𝑃𝑃𝑃𝑏𝑏 Eq. 3-8
The coefficients of determination (R2) associated with these correlations ranged from
0.87 to 0.99. Figure 3-7b shows the extrapolation of Poison’s ratio to 2100 psi for well GML-2.
Young’s moduli and Poisson’s ratios obtained from the above analyses are shown in
Table 3-4.
Table 3-4 Static Young’s moduli (E) and Poisson’s ratios (v) estimated at confining pressure equal to 2100 Psi.
Well
Name
Rock
Type
E
(psi)
v Well
Name
Rock
Type
E
(psi)
v
GMK-1 Shale 21310.5 0.138 GMK-2 Sand 481587.8 0.136
GMK-1 Shale 15052.2 0.174 GMK-2 Sand 416566.4 0.098
GMK-1 Toba 994833.5 0.064 GMK-2 Sand 311834.7 0.091
GMK-1 Sand 2409548.4 0.177 GMK-2 Sand 390711.2 0.108
GMK-1 Sand 432726.9 0.204 GMK-2 Sand 646349.4 0.175
GMK-1 Sand 741658.8 0.216 GMK-2 Sand 990978.5 0.186
GMK-1 Sand 644250.9 0.212 GML-2 Sand 490890.0 0.168
(a) (b)
58
GMK-1 Sand 384769.1 0.187 GML-2 Sand 444975.3 0.137
GMK-1 Sand 557527.2 0.196 GML-2 Sand 395017.5 0.144
GMK-1 Sand 293478.0 0.174 GML-2 Sand 584124.5 0.214
GMK-1 Sand 353285.4 0.193 GML-2 Shale 241320.6 0.105
GML-2 Shale 312962.2 0.144
3.5 Dynamic Moduli
Wireline logs were run during the drilling stage of three wells in the field under
consideration. Bulk density, sonic compressional slowness and sonic shear slowness data as a
function of depth are available. Dynamic moduli are calculated from wireline log data applying
equations presented in Table 2-1. In well GML-2, wireline logs were run from 1900 to 3200 m.
Figure 3-8a shows both dynamic Young modulus and shear modulus profiles, whereas Figure
3-8b presents bulk modulus and Poisson’s ratio profiles corresponding to well GML-2.
Figure 3-8 Dynamic moduli. (a) Young’s modulus, shear modulus and compressional over shear slowness from wireline logs, (b) Bulk modulus and Poisson’s ratio profiles from
wireline logs.
0 0.2 0.4 0.6 0.81800
2000
2200
2400
2600
2800
3000
3200
3400
1800
2000
2200
2400
2600
2800
3000
3200
34000.0E+00 1.0E+06 2.0E+06 3.0E+06
Δtc/Δts
Dep
th, m
Young,s Modulus and Shear Modulus, Psi
Young's ModulusShear ModulusΔtc/Δts
0.1 0.3 0.51800
2000
2200
2400
2600
2800
3000
3200
3400
1800
2000
2200
2400
2600
2800
3000
3200
34005.0E+05 1.0E+06 1.5E+06 2.0E+06 2.5E+06 3.0E+06
Poisson's Ratio
Dep
th, m
Bulk Modulus, Psi
Bulk ModulusPoisson's Ratio
(a) (b)
59
The above profiles correspond to dynamic properties since their estimation is developed
from well logs under undrained conditions, i.e., high frequency (10 kHz) and slight strain. These
conditions lead to higher stiffness of the rocks than the values measured in the laboratory under
static loading (Qiu, 2005).
The best way to correct dynamic elastic moduli is developing both triaxial tests and sonic
wave velocity measurements on rock samples and generating relationships to correct the
continuous dynamic elastic moduli.
In order to correct the dynamic elastic moduli of GML-2, some triaxial tests were
performed and validated with tests run in two neighboring wells (GMK-1 and GMK-2).
Although, GMK’s tests were developed at different confining pressure, these were corrected
using GML’s confining pressure.
Figure 3-9a and Figure 3-9b present a comparison between static and dynamic Young’s
modulus and Poisson’s ratios, respectively.
60
3.6 Dynamic to Static Young’s Modulus Correlation
There are several empirical relations to convert dynamic moduli into static moduli.
However, the majority of them only apply to hard rocks, which require smaller Young’s modulus
corrections as compared to the ones for soft rocks. The empirical equations presented in Table
2-2 are relationships between static and dynamic elastic Young’s modulus applicable to
particular sets of conditions and types of rocks. These equations can be used to predict static
moduli when laboratory tests are not available.
Static moduli data from triaxial tests and dynamic moduli data from acoustic sonic wave
velocity measurements on rock samples for sandstones and shales are available for GML and
GMK fields.
Figure 3-9 Static and dynamic moduli. (a) Red points represent static Young’s moduli and the blue continuous line corresponds to the dynamic Young’s moduli, (b) Black points
represent static Poisson’s ratio values and the red continuous line corresponds to dynamic Poisson’s ratios.
(a) (b)
61
The following procedure is established to develop an empirical relationship to accurately
estimate static Young’s modulus from dynamic data,
1) Transform static elastic Young’s moduli of GML and GMK fields to the field
mean effective in-situ stress.
2) Build a cross-plot of dynamic versus static Young’s moduli.
3) Apply available empirical relationships to predict static data from dynamic data.
4) Compare predicted values against measured data.
Figure 3-10a shows the variety of available correlations to predict static Young’s
moduli. The graph shows that the best match to experimental data is obtained with Wang’s linear
correlation, which was developed for soft rocks, while the other correlations were developed for
hard sandstones. This is consistent with the fact that the reservoirs in GML and GMK fields are
composed of soft sandstones. The coefficient of determination obtained using Wang’s equation
is 0.62. Although not perfect, the fit provided by Wang’s equation is superior to the match
obtained from other relationships.
However, since the coefficient of determination obtained from the match of predicted and
real data is only 0.62, this thesis develops a new non-linear function to improve the prediction
accuracy of static Young’s modulus by minimizing the square of the residuals. This new
relationship is developed from 22 triaxial tests conducted in GML and GMK fields and it can be
applied to either sands or shales.
A linear regression of Edy vs. Est with an R2 equal to 0.78 is proposed. It is pointed out,
however, that this model tends to underestimate values as compared with experimental data for
dynamic Young’s modulus above 40 GPa. This linear function is mathematically expressed by
the following equation,
62
𝐸𝐸𝑠𝑠𝑠𝑠 = 0.395 ∗ 𝐸𝐸𝑑𝑑𝑦𝑦 − 1.993 Eq. 3-9
Finally, a non-linear relation between log10 Edy and log10 Est with a R2 equal to 0.89 is
established. This results in a better approach for estimating static Young’s Modulus values. The
non-linear regression is mathematically expressed as follows,
log10 𝐸𝐸𝑠𝑠𝑠𝑠 = 1.494 ∗ log10 𝐸𝐸𝑑𝑑𝑦𝑦 − 1.2355 Eq. 3-10
Where Est and Edy are expressed in GPa.
Figure 3-10b presents both linear and non-linear regression models introduced in this
work. It can be clearly observed that the static Young’s modulus is calculated with a higher
degree of accuracy. Figure 3-11 presents dynamic Young’s Moduli and converted static
Young’s Moduli as compared to the real experimental data. The yellow continuous line
represents the calculated static Young’s Modulus from Eq. 3-10. The outliers in Figure 3-11 are
due to failures during the triaxial tests.
Figure 3-10 Young’s modulus correlations. a) Available relationships vs. experimental data, b) Relationships developed in this work vs. experimental data.
0
10
20
30
40
50
60
0 10 20 30 40 50 60
Stat
ic Y
oung
's M
odul
us, G
Pa
Dynamic Young's Modulus, GPa
Canady
Eissa & Kazi Linear
Eissa & Kazi Log-Log
Morales & Marcinew
Wang's correlation
Experimental Data
0
5
10
15
20
25
30
0 10 20 30 40 50 60
Stat
ic Y
oung
's M
odul
us, G
Pa
Dynamic Young's Modulus, GPa
Experimental Data
Non-Linear Regression
(a) (b)
63
Figure 3-11 Dynamic Young’s moduli in the blue continuous line and converted static Young’s modulus in the yellow continuous line as compared to real experimental data.
3.7 Dynamic and Static Poisson’s Ratio Correlation
Poisson’s ratio (ν) is analyzed at mean effective in-situ stress conditions based on
experimental data from GML and GMK fields for both shales and sandstones. Figure 3-12a
shows a cross-plot of dynamic Poisson’s ratio (νdy) versus static Poisson’s ratio (νst). Two
different trends are observed for the sandstone reservoir, one for the GML-2 and GMK-1 data
and other for GMK-2. A straight line drawn through the data points show a low coefficient of
determination (R2 =0.019). A line is not shown for shales as there are only two points in each
field and there is no correlation between them.
Since the priority of this section is to correlate νdy and νst for the sandstone reservoir in
GML field, two actions were performed in order to order to improve the quality of the
correlation between these two variables: a) the shale data and GMK-2 sandstone data were
1800
2000
2200
2400
2600
2800
3000
3200
34000.0E+00 5.0E+05 1.0E+06 1.5E+06 2.0E+06 2.5E+06 3.0E+06 3.5E+06
Dep
th, m
Young's Modulu, Psi
Experimental
Dynamic
Static
64
removed, and b) it was assumed that the maximum value for the dynamic Poisson’s ratio is 0.5
whereas the maximum value for static Poisson ratio is 0.3. The application of the aforementioned
assumptions leads to the results shown in Figure 3-12b, where the correlation between νdy and
νst is a R2 =0.7078. This correlation is expressed by Eq. 3-11.
𝑣𝑣𝑠𝑠𝑠𝑠 = 0.476 ∗ 𝑣𝑣𝑑𝑑𝑦𝑦 + .0565 Eq. 3-11
Figure 3-12 Poisson’s ratio in sandstones. (a) Poor linear fit as a results of experimental data for sandstone of well GMK-2. (b) The linear fit improves when experimental data for
sandstone GMK-2 are removed from the analysis.
Applying Eq. 3-11, dynamic Poisson’s ratio is converted to static Poisson’s ratio as
shown in Eq. 3-11.
y = 0.4763x + 0.0565R² = 0.7078
0.0
0.1
0.2
0.3
0.4
0.5
0 0.1 0.2 0.3 0.4 0.5
Stat
ic P
oiss
on's
Rat
io
Dynamic Poisson's Ratio
SandstoneCalculated dataLinear (Sandstone)
y = -0.1354x + 0.2027R² = 0.0191
0.0
0.1
0.2
0.3
0.4
0.5
0 0.1 0.2 0.3 0.4 0.5
Stat
ic P
oiss
on's
Rat
io
Dynamic Poisson's Ratio
SandstoneShale GMK-1Shale GML-2
(a) (b)
Sandstone GMK-2
65
Figure 3-13 Conversion of dynamic Poisson’s ratios (red line) to static Poisson’s ratios (yellow line) and its comparison to experimental data.
Once Young’s modulus and Poisson’s ratio are corrected, shear modulus and bulk
modulus are calculated using Eq. 3-12 and Eq. 3-13, respectively. The same methodology is
applied to the wells located close to GML field.
3.8 UCS and Internal Friction Angle Estimation
The internal friction angle (φ) and unconfined compressive strength (UCS) are two other
important geomechanical parameters that have to be considered during the analysis of
geomechanical problems. These parameters are traditionally determined from triaxial tests on
1800
2000
2200
2400
2600
2800
3000
3200
34000 0.1 0.2 0.3 0.4 0.5
Dep
th, m
Poisson's Ratio
Experimental
Dynamic
Static
𝐺𝐺 =𝐸𝐸
2 + 2𝑣𝑣 Eq. 3-12
𝐾𝐾 = 𝐺𝐺2(1 + 𝑣𝑣)
3(1 − 2𝑣𝑣) Eq. 3-13
66
cylindrical samples of rock. The fact, however, is that these data are not always available for the
reservoir zone and much less for the overburden and underburden rocks. For this reason, several
authors have published empirical correlations for estimating φ and UCS from some other
variables such as porosity, grain volume, acoustic velocity, gamma ray and, Young’s modulus.
The φ and UCS data for the upper reservoir in GML are estimated from the interpretation
of triaxial tests. However, there is lack of data for the overburden and underburden. Because of
this the empirical correlations developed are used to predict φ and UCS both inside and outside
the reservoir.
In this work, a relationship as a function of Young’s modulus is established based on
experimental data from19 UCS measurements conducted on sandstone samples collected in
lower Miocene formation (3 from GML field and 16 from GMK field), and 4 UCS
measurements carried out on shale samples (2 from GML and 2 from GMK).
Figure 3-14 shows a cross-plot of Young’s modulus vs. UCS. Scattering of the data is
observed in this figure, However, a positive correlation can be clearly established since the larger
the value of Young’s modulus, the larger the UCS. There are no outliers affecting this
correlation, which is applicable to either sandstones or shales because it indistinctly describes
both types of rock. A coefficient of determination (R2) larger than 0.95 is obtained in this case.
Figure 3-14 also presents the empirical correlation developed by Bradford (1998) for
comparison with the empirical relation for GML and GMK developed in this thesis. The graph
shows that Bradford’s correlation overestimates the UCS in comparison with GML data.
Therefore, Eq. 3-14 is used to predict UCS values in the reservoir under consideration since the
equation is valid for soft sandstones with relatively low UCS, which is the case of the reservoir
considered in this study.
67
𝑈𝑈𝑈𝑈𝛿𝛿 = 3.1081 ∗ 𝐸𝐸 + 1.44 Eq. 3-14
Figure 3-14 Young’s modulus versus UCS. The internal friction angle (φ) of a rock is barely unique since the strength of the rock is a
function of confining pressure, and also because the failure envelope in the Mohr-Coulomb
diagram is not always a straight line. For this reason, the internal friction angle is obtained from
the best straight line to Mohr circles. Due to the simplicity of the aforementioned method for
estimating φ, some authors have tried to correlate φ to porosity, acoustic velocity and gamma ray
logs.
Perkins and Weingarten (1988) observed that there is a strong correlation between
porosity and internal fiction angle. Instead of calculating internal friction angle from Mohr-
Coulomb envelope as the common practice dictates, they calculated it from the angles of failure
planes of samples with a wide range of porosity. Perkins and Weingarten based their correlation
y = 3.1081x + 1.4476R² = 0.9536
y = 3.9736x + 3.3302R² = 0.9621
0
10
20
30
40
50
60
70
80
0 2 4 6 8 10 12 14 16 18
UC
S, M
Pa
Young's Modulus, GPa
Unconfined Compressive Strength
Sandstone GMKShaleSandstone GMLBrandford's RelationLinear (Sandstone GMK)Linear (Shale)
68
on their own experimental data and data available in the literature. Their correlation for
estimating the internal friction angle (φ) as a function of porosity (ϕ) is expressed by Eq. 3-15.
𝜑𝜑 = 57.75 − 1.05∅ Eq. 3-15
In order to verify the validity of Perkins and Weingarten equation for GML field, a
crossplot of ϕ versus φ was built based on the information of 14 pairs of ϕ and φ data measured
on sandstones samples collected in the lower Miocene formation in GMK Field. Unfortunately,
porosity values were not reported for the triaxial tests conducted on samples of GML field so this
procedure could not be applied to this field.
Figure 3-15 shows a cross-plot of ϕ vs. φ for GMK field. The blue points represent
measured data and the orange straight-line represents Perkins and Weingarten empirical
correlation. Perkins and Weingarten correlation does not properly reproduce the real data and it
overestimates the values of φ. Therefore, an alternative is to draw a best fit straight line through
the GMK data. A negative correlation can be clearly established since at larger values of
porosity, lower values of φ are obtained. There are not outliers affecting the correlation. Thus the
empirical relation proposed in this work for estimating φ from porosity for soft sandstones is
expressed in Eq. 3-16.
𝜑𝜑 = 44.86 − 67.42∅ Eq. 3-16
The coefficient of determination associated with the empirical correlation introduced in
this work is low (R2=0.48), which indicates that results should be considered carefully and
interpreted as directionally correct approximations. However, this correlation represents better
the experimental data as compared with Perkins and Weingarten relation.
69
Figure 3-15 Porosity versus internal friction angle. Since, there is not static information in the overburden, sideburden and underburden for
GML field for estimating the geomechanical properties in the Mechanical Earth Model (MEM),
Eq. 3-10, Eq. 3-11, Eq. 3-12, Eq. 3-13, Eq. 3-14, Eq. 3-16 are for estimating the static
mechanical properties of the rock in the MEM from dynamic data.
y = -67.419x + 44.866R² = 0.4796
10
15
20
25
30
35
40
45
50
0 0.05 0.1 0.15 0.2 0.25 0.3
Inte
rnal
fric
tion
angl
e (d
egre
es)
Porosity (fraction)
Data GMKPerkins & Weingarten relationLinear (Data GMK)
70
Numerical Reservoir Simulation Model
Geomechanics and fluid flow coupling for field performance evaluation requires two basic
models: 1) a reservoir model that is described in this chapter for the field under study and 2) a
geomechanical model that is covered in Chapter Five. The reservoir model corresponds to a
single porosity laminated sandstone with intercalations of shale. The fluid is natural gas which is
mainly composed to methane. As a result a multiphase black oil model (gas, oil and water) is
selected for the study.
4.1 Structural Grid
Construction of the reservoir numerical simulation model was developed using Petrel and
Eclipse software. The effort started with interpretation of seismic surfaces using amplitude
attributes and continued with the match of these surfaces to well tops identified during drilling of
wells GML-1, GML-2 and GML-3. The structural grid was built considering seismic surfaces as
a guide, faults’ azimuths and dips as trends and flat spots observed in the reflectivity seismic
attribute Mu Rho (μρ) as the limit around the reservoir. Figure 4-1a shows the match between
seismic surfaces and well tops.
The grid's specifications are as follows:
• Grid orientation: 67 degrees.
• Faults and boundary edge as type zigzag.
• Block size in both i and j directions 50 m.
• Block size in k direction 1 m.
Figure 4-1b presents the grid skeleton with the three principal planes utilized to build the
3D grid as a way to honour the faults found in the reservoir.
71
Figure 4-1 Reservoir model. (a) Seismic surface and well tops. (b) 3D view of reservoir grid skeleton.
The structural grid was divided into three zones in the vertical direction; the upper
reservoir, the lower reservoir, and the intermediate shale layer, which represents the seal between
the two reservoirs. Each zone was subdivided into layers; 40 layers in the upper reservoir, 20 in
the lower reservoir, and 1 layer for the seal zone. The number of layers in each zone was
determined based on vertical thicknesses observed in the lithofacies log indicating that the
reservoirs had laminar characteristic. The reservoir grid has 171 blocks in I direction, 291 blocks
in J direction and 61 layers; therefore the total number of blocks is 3,035,421.
Three geometric attributes were considered for assessing the quality of the grid: cell
volume attribute, cell inside-out attribute and cell angle attribute. The first attribute is used to
identify blocks with negative volume whereas the second one indicates when a cell is good if its
value is equal to zero. The third attribute measures the internal angle of a corner referred to the IJ
plane and represents the deviation from 90 degrees. A cell with an angle equal to 0 degrees is a
regular block. Both cell angle and cell inside-out attributes determine the degree of deformation
of the blocks.
(b)(a)GML-3 GML-1 GML-2
72
The histogram corresponding to the cell volume attribute is shown in Figure 4-2a, which
indicates that there are not negative values in the structural grid. The minimum value of this
histogram is 1000 m3 whereas the maximum is 9750 m3. Figure 4-2b shows a histogram
corresponding to the cell angle attribute. It is noticed that more than 96% of the cells have values
smaller than 12 degrees. The maximum cell angles for this grid are approximately equal to 45
degrees and are located in the region associated with faults and at the boundary edge. The cell
inside-out attribute resulted in values equal to zero for all cells.
Based on the above results, it is possible to state that the geometry of the structural grid is
acceptable as it did not present either negative volumes or severe deformation of the blocks.
Figure 4-2 Grid quality control. (a) Cell volume attribute and resulting histogram. (b) Cell angle attribute and resulting histogram indicating good geometry in the majority of the
cells.
(a) (b)
73
4.2 Facies Modeling
Ideally the modelling of facies in numerical reservoir simulation should include
information from a large number of wells with enough core analyses and wireline logs for proper
determination of facies in the vertical direction and for performing 3D geostatistical estimations
by the use of kriging or co-kriging methods. However, due to the lack of information in the study
field (only three wells) a facies modelling process combined with a stochastic method was
applied to conceptualize the 3D facies model.
The process of sedimentary facies modeling and lithofacies modeling considers two
steps; 1) development of the vertical interpretation and, 2) areal estimation. During the first step,
sedimentary facies logs and lithofacies logs are created using information from cores and
geophysical logs. As a result of the implementation of this first step, three types of sedimentary
facies are defined: channel levee, channel margin, and channel fill, while five lithofacies are
defined as fine sandstone, coarse sandstone, shale, silt and toba.
The second step (areal estimation) considers seismic interpretation using spectral
decomposition techniques for creating maps with high resolution for the channels in the
reservoirs as shown in Figure 4-3a. Then, a correlation and match of facies is performed with
the seismic attribute μρ and porosity logs for properly defining pore fluid zones and for
delineating the channel’s polygons. The channels bodies and levees zones are defined with
seismic attributes and trends for each level and sublevel of the reservoirs as shown in Figure
4-3b.
74
Figure 4-3 Facies modeling process.
Spectral Decomposition
Match with seismic inversion Facies well log
(a)GML-2
Ampl
itud
c o
Espe
sor
Canal Margen
Método: Object Modeling
Wavelength
Am
plitu
de
Channel Levee
WidthWidthChannel Modeling
Thic
knes
s
(b)
75
The 3D sedimentary facies models are developed for each zone using well logs, channel
polygons, and probability distributions for estimating orientation, amplitude, wavelength, width
and thickness for both levees and channels. The upper reservoir is characterized by 5 different
facies bodies and the lower reservoir by 8 different facies. Each body has a specific percentage
of sedimentary facies. Figure 4-4 shows a 3D sedimentary facies distribution plot resulting from
the stochastic process and sedimentary facies modeling.
Figure 4-4 3D Sedimentary facies model.
The lithofacies model is simultaneously characterized by zonal and sedimentary facies.
Each lithofacie in the 3D model is estimated using the sequential indicator simulation method.
According to the zone and sedimentary facies, a variogram and percentage of the lithofacie is
assigned to rule the distribution and the anisotropy in both vertical and horizontal directions. The
initial percentage of each lithofacie is taken from the scaled up cells. Figure 4-5 shows a 3D
sedimentary facies distribution resulting from the stochastic process and sedimentary facies
modeling.
SedimentaryFacies
Channel Margin
Channel Levee
Channel Fill GML-2
GML-1
GML-3
76
Figure 4-5 3D lithofacies model. 4.3 Petrophysical Modeling
The procedure for estimating grid petrophysical properties is summarized in the
following steps:
1) Well logs scale-up process. During this process petrophysical properties are scaled up
from well logs to cells crossed by the well’s trajectory and a value is assigned to each
block. Properties such as porosity, water saturation, shale volume, permeability and
lithofacies are considered primary variables when applying stochastic methods for
estimating properties for the whole grid. Figure 4-6 presents effective porosity versus
depth for two wells in the field of study. The graph shows the geophysical well log to
the left and the scaled up grid cells crossed by the well trajectory to the right. An
arithmetic average method is used for scaling up porosity. The log is made up of
Lithofacies
Sand
Toba
Fine sandstone
Shale
SiltGML-2
GML-1
GML-3
77
points, which implies that the sample data within each block are used for the
averaging. All neighbouring cells around the main cell to be scaled are averaged.
The same process is followed for other petrophysical properties. The settings used
during the scale-up process are presented in Table 4-1.
Table 4-1 Well logs scale-up process and corresponding settings.
Property Average Method Treat log Method Porosity Arithmetic As points Neighbour cell Sedimentary facies Most of As lines Neighbour cell Lithofacies Most of As lines Neighbour cell Vclay Arithmetic As points Neighbour cell Permeability Geometric As points Neighbour cell Water saturation Arithmetic As points Neighbour cell
Figure 4-6 Scaled-up effective porosity (phie).
GML-2GML-1
m m
78
2) Areal estimation. Data analysis by zone and lithofacies is performed for each
petrophysical property (for example porosity, water saturation, shale volume,
permeability, NTG). Variograms are created for determining anisotropy in vertical
and horizontal directions. Porosity estimation uses the seismic porosity attribute as a
secondary variable to perform co-kriging. Other properties are calculated in the same
fashion, with the only difference that the secondary variable for the rest of the
properties is the porosity estimated previously. Figure 4-7 shows two histograms that
compare porosity and permeability distributions from three different sources utilized
for the scaling-up process and the construction of the stochastic 3D model. Although
there are small differences the comparisons are reasonable indicating a good
representation of the petrophysical properties.
Figure 4-7 Porosity and permeability histograms.
Porosity (Fraction) Permeability (mD)
Freq
uenc
y (%
)
Freq
uenc
y (%
)
79
4.4 Rock Type Definition
Aguilera (2003) indicated that reservoir porosity type can be classified as a function of
pore size and pore geometry. The pore size can be characterized by means of the pore throat
radius using different methods: Aguilera rp35, Winland r35, and Pittman rapex, where the
subscripts 35 and apex indicate 35% mercury saturation during capillary pressure estimation. For
r35≥10 microns the pore sizes are classified as megapores, for 2≤r35<10 microns as macropores,
for 0.5≤r35<2 microns as mesopores and for r35<0.5 microns as micropores.
Pittman (1992), based on 202 uncorrected air permeability and porosity analyses on
sandstone core samples developed Eq. 4-1 for determining the pore aperture size distribution
corresponding to 40% mercury saturation.
log 𝑟𝑟40 = 0.360 + 0.582 log𝐾𝐾 − 0.680 log∅ Eq. 4-1
where K is uncorrected air permeability in mD, ϕ is porosity in percentage and r40 is the pore
throat aperture in microns at 40% mercury saturation during a capillary pressure test.
Based on matrix porosity and matrix permeability data from core samples, a semi-log
crossplot of porosity versus permeability is created as shown in Figure 4-8. The dashed lines in
this figure represent Pittman r40 pore throat apertures.
According to the pore throat aperture, the analysis indicates that five rock types (TR)
are present in each reservoir.
• Rock type 1, TR1 ≥ 10μ
• Rock type 2, 2μ ≤ RT 2 < 10μ
• Rock type 3, 0.5μ ≤ RT 3 < 2μ
• Rock type 4, 0.1μ ≤ RT 4 < 0.5μ
• Rock type 5, TR 5 < 0.1μ
80
4.5 Reservoir Limits
The vertical and horizontal limits of the reservoir are defined as follows: the top limit is
the top surface of each reservoir, which corresponds to the base of the seal rock. For the upper
reservoir the top is found at a depth of 3048 m whereas for the lower reservoir it is at 3186 m.
The horizontal limits around the reservoir are defined by the intersection between the top surface
dip and the gas-water contact. At the northern part of the reservoir, the horizontal limit is defined
by a sealed fault. The gas-water contact was estimated using pressure gradient data acquired
during drilling of well GML-2 and corroborated with well GML-3. The gas-water contact for the
upper reservoir is at a depth of 3119 m and for lower reservoir at 3214 m.
Based on the pore volume located below the gas-water contact, the aquifer volume is
estimated to be 1.2 times larger than the pore volume of the upper reservoir and 2 times larger
than the pore volume of the lower reservoir. The aquifer pressure support effect on the reservoir
Figure 4-8 Rock type definition using Pittman r40. (a) Upper reservoir. (b) Lower reservoir.
81
was insufficient during the drawdown tests. The numerical aquifer included in the reservoir
model to add energy for matching the drawdown test is shown in Figure 4-9.
Figure 4-9 Aquifers associated with the upper and lower reservoirs.
4.6 Saturation Functions
Capillary pressure (Pcap) and relative permeability curves are defined for each rock
type in order to establish fluid distribution, initial water saturation and fluid flow in the porous
media. A total of 10 of capillary pressure sets and relative permeability curves are defined for the
reservoir.
4.6.1 Capillary pressure
Experimental capillary pressure data for each rock type were obtained from core
samples. However the responses for each type of rock are not unique since they are a function of
porosity, pore radius and interfacial tension. Moreover, data coming from core samples represent
a very small fraction of the reservoir. Hence, it is necessary to use functions for scaling-up the
data from a core scale to a reservoir scale. The Leverett J-function is a dimensionless saturation
function that attempts to extrapolate capillary pressure data from a specific rock sample to rocks
in the reservoir presenting different porosity, permeability and wetting properties (Kantzas at al.,
2012).
82
The Leverett J-function defined in Eq. 4-2 integrates different capillary pressures into
a common curve. Thus, it is possible to scale-up the capillary pressure for different porosity and
permeability values. Leverett J-function is mathematically expressed as follows,
𝐽𝐽(𝛿𝛿𝑤𝑤) = 𝑐𝑐𝑃𝑃𝑃𝑃𝑚𝑚𝑝𝑝𝜎𝜎𝑠𝑠
�𝐾𝐾∅
Eq. 4-2
where c is a constant, K is permeability, φ is porosity, Pcap is capillary pressure and σt is
interfacial tension.
Figure 4-10a presents an example of a J-function versus water saturation curve and its
corresponding capillary pressure curve scaled-up for rock type 1 using a porosity of 0.24 and a
permeability of 100 mD. These values of porosity and permeability are average values for a
specific rock type.
4.6.2 Relative permeability
Empirical correlations are often required for estimating relative permeability values,
especially in reservoirs where information is scarce due to technical and/or economical
limitations. One of the most important methods for estimating relative permeability data is the
Corey relationship (Brooks and Corey, 1964) presented in Eq. 4-3, Eq. 4-4 and Eq. 4-5. Two
important parameters have to be considered for relative permeability estimation: the end-points
and the curvature.
𝛿𝛿𝑤𝑤∗ =𝛿𝛿𝑤𝑤 − 𝛿𝛿𝑤𝑤𝑚𝑚1 − 𝛿𝛿𝑤𝑤𝑚𝑚
Eq. 4-3
𝐾𝐾𝑟𝑟𝑤𝑤 = (𝛿𝛿𝑤𝑤∗)𝜓𝜓 Eq. 4-4
83
𝐾𝐾𝑟𝑟𝑚𝑚𝑤𝑤 = (1 − 𝛿𝛿𝑤𝑤∗)2(1− 𝛿𝛿𝑤𝑤∗𝜔𝜔) Eq. 4-5
𝐾𝐾𝑟𝑟𝑚𝑚𝑤𝑤∗ =𝐾𝐾𝑟𝑟𝑔𝑔
�𝐾𝐾𝑟𝑟𝑔𝑔�𝑆𝑆𝑤𝑤𝑤𝑤 Eq. 4-6
𝐾𝐾𝑟𝑟𝑤𝑤∗ =𝐾𝐾𝑟𝑟𝑤𝑤
(𝐾𝐾𝑟𝑟𝑤𝑤)𝑆𝑆𝑓𝑓𝑛𝑛𝑤𝑤 Eq. 4-7
Where:
ψ = (2+3λ)/λ
ω = (2+λ)/λ
λ = Pore size distribution.
Sw* = Effective saturation.
Swi = Irreducible water saturation.
Krw = Water relative permeability.
Krnw = Non-wetting relative permeability.
Srnw = Residual non-wetting phase saturation.
Krnw* = Normalized non-wetting phase relative permeability.
Krw* = Normalized water relative permeability.
The relative permeability analysis for the reservoirs under study starts with
normalization of the curves available for each rock type by applying Eq. 4-3, Eq. 4-6 and Eq.
4-7. Next the average relative permeability curves are reproduced with Corey expressions using
exponents as tuning parameters to match the curvature. Finally, the normalized curves are
converted to relative permeability curves using average end-points (Swi and Srn w) for each rock
84
type. During each realization of the stochastic process to estimate porosity and absolute
permeability, the end-points change. Figure 4-10b is an example of relative permeability curves
for rock type 1 using in the model. Relative permeability curves for other rock types are
estimated following the same methodology described above.
Figure 4-10 Saturation functions. (a) Leverett J-function and capillary pressure for rock type 1. (b) Relative permeability for wetting and non-wetting phases for rock type 1.
4.7 Fluid Characterization
The characterization of reservoir fluids by means of PVT analysis and cubic equations of
state (EOS) allows estimating fluid properties and phase behaviour as a function of pressure,
volume and temperature. Phase behaviour is required for numerical reservoir simulation and
design of production facilities.
Bottom-hole samples and surface samples were recovered during completion of wells
GML-1 and GML-2. Both samples were analyzed using a chromatographic analyzer, which
indicated that the fluids in both reservoirs correspond to dry gas. As mentioned in Chapter One,
the reservoir fluid presents 0.0% mole H2S, 0.09-0.13% mole carbon dioxide. Dew point
pressure is equal to 383 kg/cm2, relative gas density is 0.59 and gas oil ratio is 3.9 bbl/MMscf at
standard conditions. The fluid was analyzed using the Peng-Robinson EOS, Whitson lumping
85
method and Lee mixing rules. The total number of components after the lumping process was 25
and their mole fractions are shown in Table 4-2.
Table 4-2 Reservoir fluid composition.
Name Zi Name Zi
N2 1.02 nC7 0.087
CO2 0.026 Metil-Ciclo-Hexano 0.042
C1 95.063 Tolueno 0.028
C2 1.965 nC8 0.091
C3 0.714 Etil-Benceno 0.004
iC4 0.167 p-Xylene 0.007
nC4 0.203 o-Xylene 0.004
iC5 0.082 C9+ 0.116
nC5 0.075 C12+ 0.071
C6 0.093 C15+ 0.037
Metil-Ciclo-
Pentano
0.03
C18+ 0.024
Benceno 0.023 C23+ 0.016
Ciclo-Hexano 0.012
The saturation pressure of the new mixture is matched to the experimental data using
critical properties of the pseudo components. Figure 4-11 shows the phase behaviour diagram
before and after the match. Dew point pressure is 383 kg/cm2 and initial pressure is 368 kg/cm2.
86
Figure 4-11 Phase behaviour diagram before and after the match.
Results of the simulation of the constant composition expansion experiment (CCE) are
shown in Figure 4-12, which presents a reasonable match between the experimental data and the
simulated data. Thus, the Peng-Robinson EOS is used in the numerical reservoir simulation
model to properly represent fluid behaviour.
87
Figure 4-12 Constant composition expansion experiment at 60.5 °C. Red dots represent experimental data. The lines represent the simulated data.
In summary, GML fluid presents a high concentration of methane (95% mole) and a low
gas-oil solubility ratio (191,836 ft3/bbl), which is an indicator of a gas and condensate reservoir
but with very low content of condensate. The condensate density is 35 °API, which is low as
compared with the density of a typical gas condensate reservoir (60 °API). Also Figure 4-11
shows that reservoir pressure and temperature are within the phase envelope, which indicates that
condensation is present in the reservoir. However, reservoir pressure and temperature conditions
are above the critical point of the mixture (high concentration of methane). Therefore, the
reservoir can be classified as atypical gas reservoir with very low content of condensate.
88
4.8 Vertical Flow Performance
Gas production at surface conditions is governed by the pressure drop along the production
tubing during the lifting of fluids from the bottom of the hole to the surface. Thus, vertical flow
performance tables were used for estimating the relationship between pressure and fluid flow
rate as the reservoir goes on production. Taking into account the possibility of vertical and
directional well’s trajectories during the development of the field, tables were designed
considering vertical and directional trajectories, 5- inche choke, and tubing head pressure ranging
from 36 bar to 297 bar. Therefore, the simulated well gas production rate varies up to 3 million
m3/day as illustrated in Figure 4-13.
Figure 4-13 Vertical flow performance table (well gas production rate versus bottom hole pressure) used in well gas production simulation.
89
4.9 Development Strategy
The production strategy implanted in the reservoir was defined using the following
assumptions:
• 7 comingled producer wells (the appraisal wells are considered as development
wells).
• The well trajectory crosses both reservoirs.
• Wells are located along the top of the reservoir.
• Minimum drainage radius is 650 m.
• Maximum field gas production rate is 400 MMscfd.
• Minimum economical production rate is 50 MMscfd.
• Tubing head pressure is 80 bar along the production life of the reservoir.
Based on the assumptions mentioned above, 7 wells are defined throughout the reservoirs
as shown in Figure 4-14. Three wells will be active during the first 6 months of production with
a total rate equal to 200 MMscfd. The remaining wells will enter production after 6 months to
reach a maximum field gas production rate equal to 400 MMscfd. Both bottom-hole pressure and
tubing head pressure will govern the extension of the production plateau of the field.
90
Figure 4-14 Location of the wells throughout the reservoir under study. 4.10 Reservoir Numerical Simulation Results
The Sequential Gaussian Simulation (SGS) or any other stochastic methods such as Monte
Carlo method and Gaussian random function are conditional simulation processes used to
generate equally probable maps of properties that are within the uncertainty estimates made by
the kriging or cokriging processes (Hirsche, 1996). Consequently, SGS is used for handling
spatial uncertainty. These simulations are conducted using specialized software, which gives the
option to constrain the simulations by a histogram distribution of values; however, due to the
limited sample size in the study field this option is not used. Thus, cokriging model is chosen to
perform the stochastic simulation.
300 realizations are created using the reservoir numerical simulation model to determine a
reservoir model with 50% probability of concurrency of original gas in-place. The original gas
91
in-place volume distribution presents a standard deviation equal to 93 Bscf and the mean is 1,180
Bscf. The original gas in-place percentiles are shown in Figure 4-15a: P(10) is 998 Bscf, P(50)
1,150 Bscf and P(90) 1311 Bscf.
Considering the realization corresponding to P(50) original gas in-place, the numerical
reservoir simulation model is run under the constrains indicated in the development strategy
section. The estimated field gas production rate (FGPR) is characterized by three stages as
illustrated in Figure 4-15b. The first stage corresponds to the four wells producing a total of 200
MMscfd during the first 6 months of production life of the field. The second stage of production
is characterized by the 7 wells producing a total rate equal to 400 MMscfd during 38 months.
Finally, the third stage of production represents the decline of the production during 40 months
until the minimum economical production (50 MMscfd) is reached. The cumulative gas
production at the end of the production life of the field (FGPT) is 733 Bscf.
Figure 4-15 Reservoir simulation model results. (a) Original gas in-place histogram built from 300 realizations. (b) Field daily gas production rate and its corresponding cumulative
gas production.
(a) (b)
Original gas in-place (Bscf)
92
Mechanical Earth Model
The Mechanical Earth Model (MEM) is a small scale numerical representation of
mechanical properties and state of stress for a particular volume of rock in a field or basin. The
MEM captures geological, geophysical and mechanical properties of the rock along with the
relationship between mechanical properties and stresses (Serra et al., 2012). The MEM is
constructed in this thesis using petrel, eclipse and visage software, data from wells in the field
under study and neighboring wells, which help to estimate the mechanical properties of the rock
outside the reservoir through stochastic methods. Data includes information such as sonic,
density and gamma ray logs, and elastic moduli from core analyses. The following sections of
this chapter describe the process to construct and validate the MEM for the offshore field
considered in this study.
5.1 Grid
The construction of the 3D mechanical earth mode (3D-MEM) has as its starting point
embedding the reservoir grid by the inclusion of sideburden, underburden and overburden. The
motivation of this embedding process is to eliminate boundary effects from the computed results
at the reservoir level.
The MEM grid for the offshore field considered in this study has been built by
embedding the reservoir grid in three zones: sideburden, overburden including the water depth
and underburden, each one with specific characteristics described in the following sections.
5.1.1 The Sideburden
The cells in this zone were generated by extending the embedding reservoir grid in I and
J directions. The grid was prolonged by 36.0 km and 40.0 km at each side respectively and
subdivided into 10 geometrically spaced cells with a factor equal to 1.3. There are two reasons
93
for these specific distances in I and J directions. First, it is recommended for the model to have a
square areal shape (Schlumberger Information Solutions, 2011); second, the MEM grid has to be
aerially larger (it is recommended between 3 and 5 times) than the unembedded grid.
Additionally, a stiff plate was considered and 50 m cells were added around the sideburden to
guarantee uniform applications of loads to the embedded grid. The final areal extension is 84 km
by 84 km. The rotation angle is 67 degrees. Figure 5-1 shows the cell distribution in the
sideburden zone.
Figure 5-1 Sideburden cells distribution considering geometrical spacing.
5.1.2 The Overburden
The study field is located offshore in the Gulf of Mexico in water depths ranging from
980 m in the southern part of the field to 1,200 m in the northern part. In order to honour these
water depth variations in the MEM, a seabed surface was created using seismic information and
well tops from 8 wells drilled in the study area. The surface is presented in Figure 5-2a. In
similar manner, a subsurface for the Upper Pliocene formation was generated as shown in
Figure 5-2b.
94
Figure 5-2 Surface maps. (a) Seabed surface configuration and (b) Upper Pliocene surface.
(a)
(b)
95
The overburden cells were built using the previously mentioned surfaces and the top
surface of the upper reservoirs. The distance between the seabed surface and the upper Pliocene
surface was divided in 2 proportional layers with equal sizes and the distance between the upper
Pliocene surface and the top surface of upper reservoir was split in 20 layers following a
geometric progression. In order to make the layer with a similar size to the corresponding layer
in the reservoir, a geometric progression factor of 1.2 was defined. Figure 5-3 shows the layers
distribution in the overburden.
Figure 5-3 Overburden cells distribution. 5.1.3 The Underburden
The underburden was divided into three zones for layering purposes. The first zone was
divided in 7 layers using the geometric progression method and a factor equal to 1.5. These 7
layers go from the base of the lower reservoir to a depth of 5500 m. There is well log data for
some wells that go down to this depth. The second zone was defined from 5,500 m to 17,000 m
and was split into 5 layers using a geometric progression method and a factor equal to 1.5. The
third zone was defined from 17,000 m to 29,000 m. with 4 layers of equal sizes. The reason for
this final depth is because the aspect ratio in the model is recommended to be greater than 3:1
and the thickness of each layer should be approximately the depth of the reservoir’s base
including the water depth (Schlumberger Information Solutions, 2011). For example, 6 km of
96
horizontal extension requires at least 2 km of depth and a reservoir’s base at 3,000 m implies 4
layers with thickness of 3,000 m. Figure 5-4 presents the underburden distribution.
Figure 5-4 Underburden cells distribution.
5.1.4 MEM grid
The final MEM grid is a cube with dimensions of 84 km x 84 km x 29 km. The total
number of cells is 6,239,376 cells, with 303 cells in I direction, 208 cell in J direction and 99
layers. Figure 5-5 is shown the final MEM grid with all the specifications previously mentioned.
Figure 5-5 MEM 3D grid. Reservoir, underburden, overburden and sideburden.
Sideburden
Embedded Reservoir
Underburden
Overburden
97
5.2 Grid Quality Control
This is an important phase during grid construction and helps to guarantee that the grid is
usable for the MEM. The quality control helps to meet the following simple conditions.
Cell volume: The block volume must be bigger than zero and negative values will
produce errors. The most common cause of negative volume is associated with faults that cross
each other. The lowest cell volume in this study is 1,363 m3. Therefore, the grid covers the
requirement of no negative volumes.
Gaps: The grid must not include gaps. Although the reservoir grid contains a shale layer
between the two reservoirs that is a seal for the lower reservoir, the later was considered to be
active and part of the MEM to avoid gaps in the grid.
Cell inside-out: This property measures the quality of a grid block. When the cell inside-
out value of a grid block is equal to zero, the block has good geometry; otherwise the cell has a
certain degree of distortion and poor geometry. All MEM grid cells developed in this study have
values of cell inside-out equal to zero.
Pinch-outs: They occur when two corners have the same coordinate. The 3D-MEM has
no pinch-out problems.
Cell angle: This property represents a deviation from the 90 degrees angle reference. In
this thesis, the IJ plane was used to extract the internal angle of deviation for every cell and the
results indicate that there are some cells with angles differences between 30 and 48 degrees
(maximum). Since the number of cells with high angle is small (798 cells), just 0.02% of total
cells, it is considered that these cells do not generate a convergence problem during the
simulation.
98
5.3 Single Material Test
The first step for evaluating the suitability of the grid geometry for a gravity/pressure
stress state analysis is the creation of a single material MEM with average properties to analyse
the total stresses at zero time step and elastic conditions. The data presented in Table 5-1 were
the input to the single material MEM. This considers sandstone with isotropic and intact rock
material.
Table 5-1 MEM single material properties.
Property Value
Young’s Modulus = 10 GPa
Poisson’s Ratio = 0.3
Bulk density = 2.3 g/cm3
Biot Elastic constant = 1
Termal Expansion Coeff = 1.3E-05 1/°K
Porosity = 0.3
The initial conditions for application of the gravity/pressure method were defined as
follows:
• Minimum horizontal gradient= 0.20 bar/m = 20 KPa/m
• Shmin offset= 0
• SHmax/Shmin = 1.2
• Sea fluid pressure gradient= 0.101 bar/m = 10.1 KPa/m
• Pressure of undefined cells= 0 bar/m = 0 KPa/m
• Offset= 0
• Minimum horizontal stress Azimuth = 0 degree
99
5.3.1 Single material test results
The cube shape and single material characteristics allow boundary stresses to be applied
uniformly to the model where the faulting regime is assumed normal. The vertical gradient input,
based on bulk density, was 0.225 bar/m (22.5 KPa/m), whereas in the model it is 0.229 bar/m
(22.9 KPa/m). Likewise, the minimum horizontal stress gradient input was 0.20 bar/m (20
KPa/m), while in the model the minimum horizontal stress, acting in the “y” direction and its
gradient is 0.192 bar/m (19.2 KPa/m). The maximum horizontal stress gradient in the model is
acting in “x” direction and is equal to 0.24 bar/m (24 KPa/m). Thus, the resulting ratio
SHmax/Shmin in the model is 1.25, which is slightly larger than the input ratio (1.20).
Figure 5-6 shows a cross plot of stresses extracted at GML-2 location vs. depth, where
TOTSTRXX is the total stress in the x direction, TOTSTRYY is the total stress in y direction
and TOTSTRZZ is the total stress in z direction. All of them exhibit a linear gradient. Hence, it
can be concluded that the MEM grid is working well and applies correctly the boundary stresses
to the model.
100
Figure 5-6 Single material test using the MEM 3D grid. 5.4 Mechanical Properties Extrapolation
There are different ways of assigning mechanical properties to the MEM grid. For
example the properties can be transferred from one grid to another grid, or calculated from well
logs and extrapolated to the entire grid using kriging interpolation method, or co-kriging in
101
combination with stochastic methods. Other way to do it is using the graded method which is
explained next.
For the offshore reservoir under consideration the MEM grid geomechanical properties of
the overburden and sideburden were estimated with the use of the following two steps:
1) Scale-up from well log to well grid cells.
2) Populate the whole MEM grid using stochastic methods.
The first step is related to the vertical property estimation in the grid cells that pass-
through the well trajectory. It was developed using the settings indicated in Table 5-2.
Table 5-2 Settings for scale-up of properties from well logs
Scale-up property Average Method Treat log Method
Porosity Arithmetic As point Neighbor cell
Young’s Modulus Mid-point As point Neighbor cell
Poisson’s ratio Maximum As point Neighbor cell
UCS RSM As point Through cell
Bulk density Median As point Neighbor cell
Friction RMS As point Neighbor cell
The average method selected to scale-up the well logs was based on the match between
the log property and the final average property in the grid cells. Figure 5-7a to Figure 5-7d
show histograms for some of the properties listed in the previous table. The graphs show good
matches indicating that the data are represented properly in the MEM.
102
Figure 5-7 Scale-up properties from well logs, well log in solid pink color, scale-up cells in solid green color. (a) Porosity, (b) Young’s modulus, (c) Poisson’s ratio and (d) UCS.
The second step considers the areal and vertical estimation. According to Isaak and
Srivastava (1989), the estimation of a variable when there is no sample requires a model of how
the variable behaves in that specific area. Without a model it is impossible to estimate a value in
(a) (b)
(c) (d)
Porosity (Fraction)
Freq
uenc
y (%
)
Young’s Modulus (Gpa)Fr
eque
ncy
(%)
Poisson’s ratio (Fraction)
Freq
uenc
y (%
)
Freq
uenc
y (%
)
UCS (MPa)
103
a location that was not sampled. Base on the quality of input data, petrophysical modeling
process allows estimating values of a property in a location that has not been sampled, using
either deterministic or probabilistic methods.
The deterministic methods will always predict the same result with the same input data.
The most common deterministic method is Kriging. In contrast, the probabilistic or stochastic
methods predict different results with the same input data.
The stochastic (probabilistic) simulation is the action of producing different models with
the same probability of occurrence in a random field. There are several methods for creating
stochastic models. However, the most common is Sequential Gaussian simulation, which must
satisfy three requirements: data values at specific location, histogram with normal distribution
and random function model (variogram).
The field MEM property modeling was performed by vertical zones (7 vertical zones).
However not all the zones were modeled. These vertical zones were defined as follow:
• Zone 1, from seabed to upper Pliocene top
• Zone 2, from upper Pliocene to upper reservoir top
• Zone 3, from upper reservoir top to Upper reservoir base
• Zone 4, from upper reservoir base to lower reservoir top
• Zone 5, from lower reservoir top to lower reservoir base
• Zone 6, from lower reservoir base to -5,500 m.
• Zone 7, from 5,500 m. to a depth -29,000 m.
The last zone was not modeled using deterministic nor stochastic methods because there
are not well logs to scale-up. Instead, this zone was populated using the graded method explained
next.
104
Due to lack of areal sample data (only 6 points), it was not possible to apply any
deterministic method such as kriging, and instead the Sequential Gaussian simulation method
was applied for estimating values of geomechanical properties in the MEM.
In general the properties do not present normal distributions. Therefore, transformation of
input distribution into normal distribution for each vertical zone was carried out with 2 different
methods: the Normal score and the Beta transformation functions. Then the data are back-
transformed,
The normal score transformation converts the input data into a standard normal
distribution with the use of Blom’s equation (Eq. 5-1) (“Normal Score”, n.d.). Thus, each cell
value in the property domain has a value in the normal score domain through the same
cumulative probability in both domains. This type of transformation is not recommended when
there are few input data points available because the resulting histogram will be poor of
resolution.
𝑠𝑠 = 𝜂𝜂 �𝑟𝑟𝑎𝑎 − 3
8𝐿𝐿 − 1
4� Eq. 5-1
Where, s is the normal score for an observation, ra is the rank for that observation, n is
the sample size and η is the point quartile from the standard normal distribution (cdf).
The Beta distribution transformation assumes that the input distribution can be matched
with a beta distribution function and then transformed it into a standard normal distribution. The
probability density function is expressed by equation Eq. 5-2.
𝑓𝑓(𝑥𝑥;𝛼𝛼1,𝛽𝛽1) =𝑥𝑥𝛼𝛼−1(1 − 𝑥𝑥)𝛽𝛽−1
𝐵𝐵(𝛼𝛼1,𝛽𝛽1) Eq. 5-2
105
where B(α1, β1) is the Beta function that is normalized to ensure the total probability is equal to
1. α1 and β1 are defined as function of variance (var) and mean (�̅�𝑥) are defined as shown in
equations Eq. 5-3 and Eq. 5-5 and η is a positive integer (Soong, 2004).
𝛼𝛼1 = �̅�𝑥 ��̅�𝑥(1 − �̅�𝑥)𝑣𝑣𝑎𝑎𝑟𝑟
− 1� Eq. 5-3
𝛽𝛽1 = (1 − �̅�𝑥)��̅�𝑥(1 − �̅�𝑥)𝑣𝑣𝑎𝑎𝑟𝑟
− 1� Eq. 5-4
𝐵𝐵(𝛼𝛼1,𝛽𝛽1) =Г(𝛼𝛼1)Г(𝛽𝛽1)Г(𝛼𝛼1,𝛽𝛽1) ; Г(𝜂𝜂1) = (𝜂𝜂1 − 1)! Eq. 5-5
If an input histogram exhibits a Beta distribution function, the algorithm transforms it
closely to a Gaussian distribution. Thus, the more similar the data distribution to the Beta
function the better the output accuracy in the model. The data analysis for the distribution
properties in the overburden is shown in Appendix A.
The stiff plate is modeled elastically and it is recommended to have a Young’s modulus
that could be twice the mean value or 50% of the pick; further it is recommended to have
Poisson’s ratio corresponding to a low value of the data (Schlumberger Information Solution,
2011). For bulk density and the UCS it is recommended to use the highest value of the data. The
values so defined for the stiff plate are shown in Table 5-3. The material type selected is intact
rock material with isotropic elastic model and Mohr-Coulomb yield criteria.
106
Table 5-3 Stiff plate properties.
Property Value Property Value
Young’s Modulus = 4.12 GPa UCS = 2.48 MPa
Poisson’s Ratio = 0.12 Friction Angle= 46 deg
Bulk density = 2.55 g/cm3 Dilation Angle = 23 deg
Biot Elastic constant = 1 Tensile Stress = 2.5 MPa
Termal Expansion Coeff = 1.3E-05 1/K Hardening/Softening Coeff = 0
Porosity = 0.01
As mentioned above the underburden was layered in three vertical zones; however, the
estimation of properties was divided into two zones: the first from the base of the embedded
reservoir to -5500 m and the second from -5500 to -2900m. Characterization of the first zone is
good because there are some well-logs. Properties were estimated using the methods described
previously. However the second zone does not have logs and as a result the grid was populated
using the graded method. The graded method is recommended when log data are not available.
The method considers a linear gradient between the values at the underburden top and
the underburden base. The value for the underburden top can be equal to the average value of the
layer just above of it or the average value observed in the reservoir. The value for the
underburden base is equal to the value used for the stiff plate. Table 5-4 presents the linear
equations used for calculating properties in the underburden using the graded method. The
underburden is modeled elastically.
107
Table 5-4 Underburden properties using graded method.
Property Linear Equation
Young’s Modulus = ( 9.447E-05 * Depth ) + 1.380
Poisson’s ratio = ( -5.489E-06 * Depth ) + 0.2792
Bulk Density = ( 1.234E-05 * Depth ) + 2.192
Porosity = ( -4.98E-06 * Depth ) + 0.1544
UCS = ( 4.596E-03 * Depth ) + 114.7
Friction angle = ( 4.255E-04 * Depth ) + 32.66
Tensile = UCS * 0.1
Dilation angle = Friction Angle * 0.5
5.5 Pre-Production Stress State
5.5.1 Minimum horizontal stress
As mentioned in Chapter 3, the most reliable and accepted method for estimating the
minimum horizontal stress is fracturing the formation and logging the closure pressure. This
closure pressure is assumed to be equal to the minimum of the three principal stresses affecting
the rock. In this model this minimum stress is considered horizontal (Barree, 2009).
Eight leak-off tests were performed in the study field; four in well GML-1, two in GML-
2 and 2 more in GML-3. Each leak-off test was analyzed using a methodology presented by
Lopez et al. (2014), based on a derivative plot of √t vs. √t*∆P/(∆√t) . Application of the
methodology is illustrated in Figure 5-8 with data of a leak-off test carried out in well GML-3 at
a depth of 2996 m. The obtained closure pressure at surface conditions is equal to 1435 psi (9.89
MPa or 98.9 bars).
108
Figure 5-8 Leak off test developed in well GML-3 at depth of 2996 m.
Table 5-5 summarizes the main results of the leak off test interpretation. Figure 5-9
presents the minimum horizontal stress gradient at reservoirs depth. The gradient varies from
0.17 to 0.20 bar/m (17-20 KPa). Minimum horizontal stress is expected to be between 439 and
506 bars at datum depth of the upper reservoir. The minimum and maximum horizontal stresses
for well GML-2 at reference depth were calculated in Chapter 3 and are equal to 452.7 bar (45.2
MPa) and 522.3 bar (52.23 MPa) respectively. Therefore, the ration SHmax/Shmin is equal to 1.15.
109
Table 5-5 Minimum horizontal stress from leak Off Test
Well Depth (m) Shmin (bar) GML-1 1440 171
GML-1 1937 268
GML-1 2665 397
GML-1 3500 601
GML-2 2273 299
GML-2 2742 379
GML-3 1996 288
GML-3 2996 475
Figure 5-9 Leak off tests carried out in offshore field under study. There are three different gradients for the minimum horizontal stress: 0.17, 0.18 and 0.19 bar/m
y = 4.81x + 655.7
y = 5.84x + 527.9y = 5.37x + 445.9
0
500
1000
1500
2000
2500
3000
3500
40000 200 400 600 800
Dep
th (m
)
Shmin (bar)
Leak Off Test GML Filed
GML-1GML-2GML-3
110
5.5.2 Stresses orientation
Well breakouts are zones that present compressive failure or wellbore enlargements. The
minimum horizontal stress azimuth is located in the zone of the wellbore wall where the stress
state is more compressive. Thus, the minimum horizontal stress azimuth is in the orientation of
the breakout (Zoback, 2006).
An easy way to identify wellbore breakouts orientation and opening angle is using the 4-
arm caliper log. This tool helps to identify breakouts by measuring the displacement of four arms
held at 90°, generally configured in such a way that opposite arms move the same amount and
assuming that the tool is always centered (Rider and Kennedy, 2011). To calculate breakouts
orientation the bit size should be known. Two independent caliper logs are measured by two
pairs of arms, resulting in two orthogonal borehole sizes. This permits calculating breakouts
from the displacements and azimuth ratio to north from one of the arms acting as an azimuth
tool.
The caliper log for well GML-1 indicates the presence of breakout zones at about 1975
m with an azimuth of approximately 120°. This was the only zone that presented clear
displacement (Figure 5-10a). Faults are also indicators of the principal stresses orientation at the
moment in which the faults were generated. Figure 5-10b shows normal faults in the field with
orientation N30°E as a consequence of convergence of stresses to the East.
111
Figure 5-10 Minimum horizontal stress orientation from caliper log and maximum horizontal stress orientation from faults.
112
Initial conditions for the pre-production stress state analysis are as follows:
• Minimum horizontal gradient= 0.15 bar/m = 15 KPa
• Shmin Offset= 0
• SHmax/Shmin = 1.1
• Minimum horizontal Stress azimuth = 120 degree
• Sea fluid pressure gradient= 0.101 bar/m = 10.1 KPa
• Pressure of undefined cells= 0.15 bar/m = 15 KPa
• Offset= 0
The model configuration was controlled using the following rules:
• Pinchout tolerance method= Factor
• Pinchout tolerance= 1.0 E-006
• Iterative solver tolerance= 1.0 E-007
• Number of increments= 4
The minimum horizontal gradient and the ratio SHmax/shmin are the result of field stress
calibrations.
5.5.3 Discontinuity modelling
The reservoir model has 8 faults. However, only the three most important faults were
considered in the geomechanics simulation for simplicity. One is the fault located in the north of
the field; this is a sealing fault and boundary for both reservoirs. The two other important faults
are in the center of the field, they do not present significant displacement and are considered
opened to flow. However, there is a possibility that these faults could be a barrier to fluid flow
113
between the side bocks. If this is the case it could have a negative impact in the cumulative gas
production. It is recommended to include all the faults in future simulation work.
In a MEM model the perfect fault to be modeled would be the fault whose extension can
be determined from seismic interpretations. This is not the case for this offshore field as faults
were only mapped between the top and base of the reservoirs. For this study they were extended
from the lower Miocene surface to a depth of -4500 m in the underburden using the same
azimuth and dip shown in Figure 5-11. The fault properties were defined using the material
library included in Visage software. Where the normal and shear stiffness are 40000 and 15000
bar/m respectively; the cohesion and tensile strength are 0.01 bars; friction and dilation angles
are 20 degrees and 10 degrees, respectively.
114
Figure 5-11 Fault extension from lower Miocene to -4500 m. (a) seismic image indicating the fault’s extension in the reservoir, (b) fault’s extension in the geomechanics model.
115
5.5.4 Pre-production stress state results (elastic run)
In similar way as described in the single material test results (section 5.3.1), a single
material model was built for gradients stress comparisons. The pre-production stresses state
analysis indicates that the input vertical gradient based on bulk density log is 0.225 bar/m (22.5
KPa/m), whereas in the model the average value is 0.216 bar/m (21.6 KPa/m). Likewise, the
input of minimum horizontal stress gradient is 0.15 bar/m (15 KPa/m), while the model it yields
an average of 0.156 bar/m (15.6 KPa/m). As opposed to the model discussed in section 5.3.1, the
minimum horizontal stress in this model is acting in the “x” direction with an azimuth of 120
degrees. The maximum horizontal stress gradient is 0.189 bar/m (18.9 KPa/m) and the ratio
Shmax/Shmin is 1.21 slightly larger than the input ratio of 1.1. The resulting vertical gradient, total
minimum horizontal stress gradient and total maximum horizontal stress gradient in the MEM
are presented in Figure 5-12.
Figure 5-12 Vertical gradient (grey marks), minimum horizontal stress gradient (blue marks) and maximum horizontal stress gradient (orange marks), (a) GML-1, (b) GML-2.
116
In order to compare the 1D MEM and the 3D MEM stresses, graphs of stress versus
depth at well location for GML-1 and GML-2 were created. Figure 5-13 presents the vertical
stress and minimum horizontal stress versus depth. In the 3D MEM the vertical stress
corresponds to the total stress in the z direction while the minimum horizontal stress matches
with the stress in the x direction.
Not always are the principal stresses aligned with either the vertical axis or the well.
Frequently they are altered as a result of stress rotation particularly around discontinuities. The
principal 1D stresses can be compared with minimum and maximum horizontal stresses in the
3D regime because of their representativeness (Schlumberger Information Solutions, 2011). The
total stress in z direction and the vertical 1D stress are nearly aligned and can be compared. In
the same way, the minimum and maximum horizontal 1D can be compared with the total stresses
in x and y directions because they are approximately aligned. The grid orientation is the same as
the maximum stress azimuth.
Examining the graphs in Figure 5-13, the conclusion is reached that the gradients defined
in the 3D model are similar and consistent with the vertical, minimum and maximum horizontal
stress especially at reservoir depth.
The study field has not gone on production yet. However as the 3D MEM has been
properly calibrated it is considered to be suitable for forecasting purposes and for performing
post production analysis.
117
Figure 5-13 1D and 3D comparison for vertical and minimum horizontal stress. (a) and (b) GML-1, (c) and (d) GML-2
118
Mechanical Earth Model Results
6.1 Stresses State Analysis
Stresses state is governed by pore pressure and three principal stresses mutually
perpendicular. The principal total horizontal stresses are not equal because the tectonic stresses
components act in different magnitude and direction. Thus, when analysing underground stresses
both principal directions and magnitudes are the main concern.
The use of a 3D mechanical earth model (MEM) to simulate the underground stress state
in a reservoir is more realistic than the use of 1D model. The 3D MEM takes into account the
spatial variation in the mechanical properties of the rock, the complexity in reservoir geometry
and the presence of discontinuities. Therefore, the one-way coupled modeling process permits
updating the stress state as function of pore pressure and fluid flow in the reservoirs. In addition,
any change in these variables can potentially affect the reservoir behaviour. The stress state in a
3D MEM is considered valid from the point of view that the initial stresses have been calibrated.
6.1.1 In-situ stresses orientation
For the field of interest, the stresses orientation in the 3D MEM coincide with the
orientation of 30 degree for the maximum horizontal stress and 120 degrees for the minimum
horizontal stress as indicated during its construction and measured by the wells. The arrows with
green head and red stem in Figure 6-1 show the calculated initial maximum horizontal stress
orientation. The minimum horizontal stress is perpendicular to the arrows.
119
Figure 6-1 XY plane showing the initial maximum horizontal stress orientation in the reservoir and close-up in blue square.
6.1.2 In-situ stress condition
The in-situ stresses determined using hydraulic fracturing and estimated through the
stress polygon indicated that the reservoir is already fractured as the stress regime for both
reservoirs is located in the transition between normal faulting and strike-slip faulting. Moreover,
the Mohr circles calculated in the 3D MEM at reservoir depth also indicate that under present in-
situ stress condition the reservoir rock has failed. This is shown in Figure 6-2 where the solid
blue line semicircle representing the initial condition of principal stresses in the reservoir has
crossed the red line representing the failure envelop of the rock.
120
Figure 6-2 Mohr circles calculated in the 3D MEM and failure envelop at reservoir depth.
The relevance of this information resides in the fact that the current conceptualization of
the porous media for these reservoirs considers only single porosity. However the above analysis
suggests that the porous media could be naturally fractured.
The preliminary evidence indicating the presence of natural fractures in these reservoirs
is found in the rock thin section slides but should be corroborated with other sources of
information such as image logs and more detailed core and petrographic work. Figure 6-3 and
Figure 6-4 are two examples that illustrate the type of fractures that could be interpreted in these
reservoirs. Figure 6-3 is a photomicrograph of a sandstone in well GML-3 at depth of 3091.3 m.
121
The petrographic study shows primary intergranular porosity with grain sizes of approximately
0.2 mm and secondary porosity developed by partial dissolution and microfractures in grains
partially cemented. Figure 6-4 is a photomicrograph from the petrographic study showing a
sandstone in GML-2 at a depth of 3068.7 m. It shows fine grains with average size of 0.07 mm
that originate primary intergranular porosity and secondary porosity developed by partial
dissolution and fractures opened and filled with organic matter.
In both examples there is a slight perception of deformation bands because of the
presence of cataclasis, dissolution and cementation as deformation mechanism. Deformation
bands are also known in the literature as gouge-filled fractures and/or granulation seams. They
occur mainly in sandstone reservoirs.
Figure 6-3 Photomicrographs of thin sections, 4x and 10x, 100 μm and 300 μm, plane light and polarized light for GML-3 at depth of 3091.3 m.
300 μm (4x) Polarized light
300 μm (4x) Polarized light
100 μm (10x) Plain light
100 μm (10x) Plain light
122
Figure 6-4 Photomicrographs of thin sections, 4x and 10x, 100 μm and 300 μm, plane light and polarized light for GML-2 at depth of 3068.7 m.
According to Fossen et al. (2007), deformation bands are limited to porous granular
media with relatively high values of porosity and encompass significant quantity of grain
translation and rotation, along with grain crushing (they are not slip surfaces). Deformation
bands occur hierarchically as individual bands and zones with band thicknesses of millimetres or
centimeters that show smaller offsets than classical slip surfaces. Deformation bands can be
classified, as shear bands, compaction bands and dilation bands (shear bands are the most widely
described in the literature) as shown in Figure 6-5 . They are characterized by four principal
300 μm (4x) Plain light
100 μm (10x) Plain light
300 μm (4x) Polarized light
100 μm (10x) Polarized light
123
deformation mechanisms: “(1) granular flow (grain boundary sliding and grain rotation); (2)
cataclasis (grain fracturing and grinding or abrasion); (3) phyllo-silicate smearing; (4) dissolution
and cementation.” (p. 757).
Figure 6-5 Kinematic classification of deformation bands. (Source: Fossen et al., 2007). Generally, deformation bands cause a decrease in porosity and permeability, which create
anisotropy and affect fluid flow. Consequently they have direct implications on management of
hydrocarbon reservoirs. Additional work is recommended previous to reaching a definitive
conclusion with respect to the presence of deformation bands.
6.2 Sequential Gaussian Simulation
Sequential Gaussian Simulation (SGS) was used to produce maps of mechanical
properties in the model with equal probability of occurrence; 10 realizations were created and all
of them honored the input data. For example, Young’s modulus average standard deviation for
the upper reservoir is 0.864 GPa and the mean is 2.131 GPa. For the lower reservoir they are
0.735 GPa and 2.547 GPa, respectively. Similarly, Poisson’s ratio average standard deviation for
the upper reservoir is 0.043 and the mean is 0.232. For the lower reservoir they are 0.064 and
124
0.222, respectively. Figure 6-6 shows two examples of SGS realizations for Young’s modulus
and Poisson’s ratio.
Figure 6-6 Two sequential Gaussian simulations realizations. (a) and (b) Young’s modulus. (c) and (d) Poisson’s ratio.
Preproduction initialization of these 10 models permits calculation of the initial stress state
according to the mechanical properties and initial pore pressure in the reservoirs. The 3D stress
state is used to verify the match between the predicted stresses and the input data. Figure 6-7
shows the distribution of principal total stresses and pore pressures at initial conditions. The
statistics for these distributions are summarized in Table 6-1. The mean values as determined in
Chapter two were Shmin=452.7 bar (45.7 MPa), SHmax=522.2 bar (52.2 MPa), Sv=534.6 bar (53.6
MPa) and Pp=358.5 bar (35.8 MPa). The difference between these values and the mean values
(a) (b)
(b) (d)
125
calculated in the model is low in spite that these properties are determined applying stochastic
methods.
Table 6-1 Statistics for distribution of minimum horizontal stress, maximum horizontal stress, vertical stress and pore pressure.
Statistic Shmin (bar) SHmax Sv Pp
Variable Upper Lower Upper Lower Upper Lower Upper Lower
Min (bar) 228.5 219.0 132.3 163.8 235.9 265.8 97.1 98.0
Max (bar) 534.3 549.9 651.8 636.3 858.5 899.5 524.6 540.5
Mean (bar) 419.0 445.6 444.6 503.8 540.6 585.5 402.8 407.3
St dev (bar) 46.3 45.46 107.8 84.78 46.7 51.5 147.4 110.7
126
Figure 6-7 Principal stresses distribution at upper reservoir datum depth. (a) Minimum horizontal stress. (b) Maximum horizontal stress. (c) Vertical stress. (d) Pore pressure.
6.3 Subsidence and Reservoir Compaction
Coupling of the models is time-consuming because the stress solution takes more time
than the flow solution when using the same grid and because the grid that considers solid
problems is at least 3 times larger the fluid flow grid. This is the reason why the coupled models
are prime candidates for high performance parallel computing.
(a) (b)
(c) (d)
127
As discussed previously in this thesis there are different ways to couple a reservoir model
with a mechanical earth model. However, previous to coupling the models it is necessary to
select the time steps at which the stress analysis will be performed in order to reduce the running
time. The most convenient way to select the time steps is using the average of field pressure
reservoir (FPR) forecast that results from the reservoir simulation model. This is used to identify
times when there is significant change in the pressure slope. In Figure 6-8 the field gas
production rate (FGPR) and FPR are plotted against time. The FPR profile does not present
strong variations. Therefore, just four stress steps were selected in order to optimize the running
time. The vertical doted red lines in the graph indicate the time step at which the stresses should
be calculated during the coupled analysis.
Figure 6-8. Time step selection to perform coupling modeling. Although the one-way coupled method is simple and has limitations compared to other
coupling methods, it is computationally efficient and accurate when reservoir compaction is
small. This method is valuable and can be applied in gas reservoir with low error in predicting
stress, strain and displacement (probably insignificant) since the rock compressibility is
considerably much lower than the gas compressibility (Gonzalez, 2012).
128
The one-way coupled method is the simplest one and considers the average pressure at
each time step for computing stress, strain and displacement. In this type of coupling method the
fluid flow equations are solved in the reservoir model and the results (pressure, temperature and
saturation) are passed on to the geomechanical simulator. However, no results are transferred
from the geomechanical simulator to the reservoir simulator.
In order to determine subsidence and reservoir compaction, the 10 models mentioned
previously were run using the one-way coupled method. Both, subsidence and upper reservoir
compaction were measured at final conditions (last day after 7 years of production). Statistical
results are summarized in Table 6-2.
Table 6-2 Univariate Statistics for subsidence and compaction
Statistic Variable Subsidence Compaction
Mean, (m) -0.7413 -1.3691
Median (m) -0.7337 -1.3413
Standard Deviation (m) 0.0730 0.2250
Variance (m2) 0.0053 0.0506
Skewness -1.207 0.258
Minimum (m) -0.9084 -1.7084
Maximum (m) -0.6396 -0.9774
Count 10 10
The univariate statistics mean for subsidence is equal to -0.7413 m and the median is -
0.7337 m. These two values are similar and could indicate a small dispersion on the data. The
standard deviation of 0.073 m indicates the dispersion of data. The coefficient of skewness is
negative and relatively large (-1.2), which means a large tail of lower values to the left of the
mean. This is observed in the histogram in Figure 6-9a.
129
The univariate statistics for upper reservoir compaction seems to present a left-skewed
bimodal distribution as shown on Figure 6-9b. The mean is 1.37 m and the median is 1.34 m, the
standard deviation is 0.22 m which represents a large dispersion.
Figure 6-9 Univariate histogram using 10 realizations. (a) Subsidence. (b) Upper reservoir compaction.
As a result of pore pressure changes in the GML reservoirs, rock collapse in the central
part of the reservoirs and significant vertical displacements are observed. The strongest depletion
occurs during the first four years when pore pressure drops 180 bars. During this period
subsidence occurs at ratio of approximately -0.003 m/bar. After the 4th year of production, the
subsidence ratio change to -0.0035 m/bar. Figure 6-10a presents the vertical displacement
distribution in the seabed at final conditions (when the reservoir reaches the economic limit). The
dark blue color in the northern part of the field highlights the zone with the largest subsidence.
Figure 6-10b shows a cross section in the j direction (AA’) that displays vertical displacement
and overburden at final conditions. The major reservoir compaction develops in the upper
reservoir and is propagated to the neighboring layers in the overburden. It is attenuated by the
overburden layer near the surface.
0%
20%
40%
60%
80%
100%
0
1
2
3
4
5
-0.9084 -0.8412 -0.774 -0.7068 -0.6397
Freq
uenc
y
Subsidence histogram
Frequency
Cumulative %
0%
20%
40%
60%
80%
100%
0
1
2
3
4
5
-1.7084 -1.52565 -1.3429 -1.16015 -0.9774Fr
eque
ncy
Compaction histogram
FrequencyCumulative %
(a) (b)
130
Figure 6-10 3D MEM rock displacement. (a) Seabed subsidence. (b) Reservoir and overburden compaction.
The upper reservoir compaction (80% of the cumulative gas production comes from this
reservoir) is characterized by two linear relationships between the vertical displacement and the
pore pressure during the production period. In chronological order, the first linear relationship
occurs at a ratio of -0.0054 m/bar during the first four years of production when the pore pressure
drops 180 bars. The second linear relationship occurs at a ratio of -0.0061 m/bar during the last 5
years of production when the pore pressure drops 83 bars. Figure 6-11 shows a crossplot of
vertical displacement against reservoir pressure. From the graph two correlations were developed
for estimating the upper reservoir compaction (ΔC) and subsidence (U) as function of reservoir
pressure.
𝑈𝑈 = 0.0031𝑃𝑃𝑝𝑝 − 1.1125 Eq. 6-1
∆𝑈𝑈 = 0.0055𝑃𝑃𝑝𝑝 − 2.0344 Eq. 6-2
A
A’
(a)A A`
(a) (b)
131
Where Pp is pore pressure in bars, U is subsidence expressed in meters and ΔC is reservoir
compaction also expressed in meters.
Figure 6-11 Verical displacement vs. reservoir pressure. Subsidence measured at seabed is shown in red. Compaction measured at the upper reservoir depth is shown in blue.
The analytical method proposed by Morita (1989) was applied to the field under study
with a view to provide a comparison with the numerical geomechanical model. Morita’s
analytical model allows quick estimation of reservoir compaction and subsidence for reservoirs
with simple geometries. The method is valuable when information is scarce as there is still a
need to make decisions on downhole and surface facility designs. Compaction in the middle of
the reservoir is estimated using Eq. 6-3 and subsidence above the reservoir center is calculated
with Eq. 6-4.
∆𝑈𝑈 = 𝑈𝑈11 + 𝑣𝑣𝑓𝑓1 + 𝑣𝑣𝑓𝑓
𝑘𝑘𝛽𝛽ℎ∆𝑃𝑃𝑝𝑝 Eq. 6-3
-1.6
-1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
100150200250300350400
Vert
ical
Dis
plac
emen
t (m
)
Pressure Reservor (bar)
SubsidenceCompaction
132
𝑈𝑈 = 𝑈𝑈3[2(1 − 𝑣𝑣𝑃𝑃)]
⎝
⎛1 −𝐷𝐷𝑟𝑟
�1 + �𝐷𝐷𝑟𝑟�⎠
⎞∆𝑐𝑐 Eq. 6-4
Where:
C1, C3: Coefficient function
Vr: Poisson’s ratio for reservoir
Vc: Poisson’s ratio for cap rock
k: Bulk modulus
β: 1-bm/Ks
Ks: Rock bulk
β: Grain compressibility
h: Reservoir thickness
Pp: Pore pressure
D: Reservoir depth
r: Reservoir radius
Subsidence and upper reservoir compaction were calculated using data presented in Table
6-3. Coefficients C1 and C3 were estimated by interpolation from graphs presented by Morita
(1989). These coefficients are functions of Log (Gr/Gc) where Gc is shear modulus for the cap
rock and Gr is shear modulus for the reservoir.
133
Table 6-3 Data required for estimating subsidence and upper reservoir compaction
Parameter Value Units
ΔPp 3769.095 psi
Er 3.80E+05 psi
Vr 0.2
Ec 2.20E+05 psi
Vc 0.3
D 10453 ft
h 232.9396325 ft
r 11365.16278 ft
ϕ 0.2 fraction
Assuming that grain compressibility approaches to zero Eq. 6-3 yields:
𝑏𝑏𝑚𝑚𝑏𝑏
~ 0
𝛽𝛽 = 1 −𝑏𝑏𝑚𝑚𝑏𝑏
= 1
log𝐺𝐺𝑓𝑓𝐺𝐺𝑃𝑃
= log𝐸𝐸𝑓𝑓/�2(1 + 𝑣𝑣𝑓𝑓)�𝐸𝐸𝑃𝑃/�2(1 + 𝑣𝑣𝑃𝑃)�
= 0.27
∆𝑈𝑈 = 0.951 + 0.171 − 0.17
∗ 1.17𝐸𝐸 − 06 ∗ 1 ∗ 232 ∗ 3769 ∗ 0.3048 = 0.386 𝑚𝑚
And applying Eq. 6-4,
𝑈𝑈 = 0.8[2(1 − 0.3)]�1 −0.9197
�1 + (0.9197)�0.386 ∗ .3048 = 0.218 𝑚𝑚
Average subsidence from the geomechanical model is -0.74 m and compares with 0.22 m
from Morita’s analytical method. Average upper reservoir compaction from the geomechanical
model is -1.37 m and compares with 0.39 m calculated from Morita’s method. The most reliable
134
results are predicted using the 3D MEM because it considers rock heterogeneity and variations in
the elastic moduli and stresses. Although there are differences Morita’s method can be used as a
first approximation if data for the MEM model are not available. The recommendation, however,
is to always collect the necessary data for building reliable MEM models. In the field under
study both the 3D MEM and Morita’s analytical method indicate that compaction and subsidence
will occur.
The production strategy for this field considers commingled exploitation of both
reservoirs using 7 subsea wells in water depths ranging from 990 to 1200 m. The plan is to install
deep water subsea tieback production facilities integrated by 2 flow lines from the field to the
onshore terminal, two manifolds, and 7 subsea vertical trees. Figure 6-12 shows a schematic
illustrating the main parts of the subsea facilities. Due to the considerable vertical displacement
observed in the 3D MEM, damage in the subsea facilities could be catastrophic if these
phenomena are not taken into account.
Figure 6-12 Schematic ilustrating the main parts of a deepwater subsea tieback production facilities.
Subsea Vertical Tree
ManifoldFlowlines
135
It is concluded that understanding and quantification of these displacement processes can
help to develop mitigation strategies to minimize and/or eliminate subsea risks. For example,
Schwall and Denney (1994) presented the case of casing deformation in the Ekofisk field
(Norway) as a result of subsidence and reservoir compaction. The consequences were axial
tubing shifts with severity ranging from slight bends to plastic tubing shear and tubing collapse.
The identification of the deformation mechanisms helped to develop techniques for minimizing
casing loading. Another case was presented by Christensen et al. (1992) who evaluated the
impact of horizontal movements related to subsidence of seabed on flowlines of the Ekofisk
field. They emphasized that these movements produced alterations in the riser conditions that
violated safety regulations; this led to the necessary equipment replacements.
As indicated above, operating companies should take into consideration the estimates of
subsidence and reservoir compaction for designing facilities and for developing mitigation
strategies that minimize or eliminate risks.
136
Conclusions and Recommendations
7.1 Conclusions
The integration of geomechanical parameters and relationships between deformation and
stress in a 3D Mechanical Earth Model (MEM) for an offshore gas field with two stacked
separate reservoirs in the Gulf of Mexico has been studied as function of changes in pore
pressure and stresses. Based on this analysis, the following conclusions have been reached:
1. New correlations have been developed to correct dynamic moduli to static conditions in
the offshore GML and GMK fields. Static and dynamic mechanical properties determined
on core samples were used for this purpose. These correlations were used to assist in
construction of a 3D MEM.
2. Average subsidence from the 3D MEM was determined to be -0.74 m. Average
compaction of the upper reservoir was determined to be -1.37 m.
3. Damage of subsea facilities could be catastrophic if subsidence and compaction
computed with the 3D MEM are not taken into account. Understanding of the
displacement processes as presented in this thesis can help to develop mitigation
strategies to minimize or eliminate risks that would damage subsurface installations.
4. The initial in-situ stress state analysis indicates that both the upper and lower reservoirs
are fractured. Their stress regime is located in the transition between normal faulting and
strike-slip faulting with a maximum horizontal stress orientation of 30 degrees.
5. Thin sections show that secondary porosity is due to partial dissolution and micro
fractures in grains partially cemented.
6. Spatial distribution of properties in the embedded reservoir; sideburden, overburden and
underburden were estimated using petrophysical parameters from the field under study
137
and neighboring fields. Geostatistical data analysis along with stochastic simulation
methods helped to generate reservoir models that honored all well data and assisted in the
uncertainty evaluation of the constructed maps.
7.2 Recommendations
1. Seismic data provides a dense sampling of the reservoir and together with petrophysical
properties and geostatistical techniques reduce the uncertainty in the distribution of rock
properties. Although the spatial distribution of properties in the reservoir models was
performed using seismic data, this was not the case for the MEM. Thus it recommended
collecting and interpreting seismic data for the whole MEM area.
2. As layers in the reservoir simulator are thin it is recommended to create a model that
coarsens up the sideburden and underburden gradually away from the reservoir.
3. Results presented in this work were developed using the one-way coupling method.
However, implementation of the two-way coupling method is recommended to handle
possible changes in porosity and permeability once the reservoir goes on production. This
approach will also provide a more rigorous production forecast.
4. Evaluate plastic deformation in order to stablish a relationship between deformation and
stresses and to analyze cap integrity.
5. Update the 3D MEM with data of new wells drilled during development of the field.
Uncertainty with respect to the future of the field will decrease as more wells are
integrated in the 3D MEM.
138
6. The mechanical property modeling has been carried out using available information but
the data bank is not considered to be complete enough for a more rigorous study. Once
new wells are drilled and more information is collected the model should be updated.
139
References
Aguilera, R. (2003, February). Geologic and engineering aspects of naturally fractured reservoirs. CSEG Recorder. Calgary, Canada. Retrieved from http://209.91.124.56/publications/recorder/2003/02feb/feb03-fractured-reservoirs.pdf
Alam, M. M. & Fabricius, I. L. (2012, January). Effective stress coefficient for uniaxial strain condition. American Rock Mechanics Association. Retrieved from https://www-onepetro-org.ezproxy.lib.ucalgary.ca/download/conference-paper/ARMA-2012-302?id=conference-paper%2FARMA-2012-302
Barree, R. D., Gilbert, J. V., & Conway, M. (2009, January). Stress and rock property profiling
for unconventional reservoir stimulation. Society of Petroleum Engineers. doi:10.2118/118703-MS
Banik, N., Koesoemadinata, A., & El‐Kaseeh, K. G. (2010). Young’s modulus from point‐
receiver surface seismic data. SEG Technical Program Expanded. 5(3), 2794–2798. doi:10.1190/1.3513424
Barton, C. A., Zoback M. D., & Burns, K. L. (1988, May). In-situ stress orientation and
magnitude at the Fenton geothermal site, New Mexico, determined from wellbore breakouts. Geophysical Research Letters. 15(5), 467-470. doi: 10.1029/GL015i005p00467
Biot, M. A. (1941, February). General theory of three-dimensional consolidation. Journal of
Applied Physics, 12(2), 155-164. Retrieved from http://www.pmi.ou.edu/Biot2005/papers/FILES/040.PDF
Biot, M. A., & Willis, D. G. (1957, December). The elastic coefficients of the theory of
consolidation. Journal of Applied Mechanics. 594-301. Retrieved from http://www.pmi.ou.edu/Biot2005/papers/FILES/068.PDF
Brace, W. F., & Byerlee, J. D. (1966). Recent experimental studies of brittle fracture of rocks.
Dept. of Geology and Geophysics, Cambridge. Mass. Brooks, R. H., & Corey A. T. (1964). Hydraulic properties of porous media. Hydrology Papers.
Colorado State University. Retrieved from http://digitool.library.colostate.edu///exlibris/dtl/d3_1/apache_media/L2V4bGlicmlzL2R0bC9kM18xL2FwYWNoZV9tZWRpYS8xNTIyNTc=.pdf
Canady, W. J. (2011, January). A Method for full-range Young’s modulus correction. Society of
Petroleum Engineers. doi:10.2118/143604-MS
140
Chang, Ch., Zoback, M.D. and Khaksar, A. (2005). Empirical relations between rock strength and physical properties in sedimentary rocks. Journal of Petroleum Science and Engineering, 51(3-4), 223-237. doi: 10.1016/j.petrol.2006.01.003
Chang, Ch., Zoback, M.D. (1998, January). Viscous Rheology and State of Stress in
Unconsolidated Sands. Society of Petroleum Engineers. doi:10.2118/47401-MS Cheng, C. H., & Johnston, D. H. (1981, January). Dynamic and static moduli. Geophysical
Research Letters, 8(1), 39–42. doi: 10.1029/GL008i001p00039 Close, D., Cho, D., Horn, F., & Edmundson, H. (2009). The sound of sonic: A historical
perspective and introduction to acoustic logging. CSEG Recorder, pp 34-43. Retrieved from http://209.91.124.56/publications/recorder/2009/05may/May2009-Sound-of-Sonic.pdf
Davis, R. O., & Selvadurai, A. P. S. (2005). Plasticity and Geomechanics. United Kingdom:
Cambridge University Press. Retrieved from https://app.knovel.com/web/toc.v/cid:kpPG000011
Detournay, E. & Cheng, A. H. D. (1993). Fundamentals of poroelasticity. Comprehensive rock
engineering, 2, 138-139. Ed. J. A. Hudson, (Pergamon Press). doi: 10.1016/0148-9062(94)90606-8
Eissa, E. A., & Kazi, A. (1988, December). Relation between static and dynamic Young’s
moduli of rocks. International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 25(6), 479–482. doi:10.1016/0148-9062(88)90987-4
FHWA. (2007). Long term pavement performance project laboratory materials testing and
handling guide. Retrieved from http://www.fhwa.dot.gov/publications/research/infrastructure/pavements/ltpp/07052/pro46/004.cfm.
Fjaer, E. (1999, January). Static and dynamic moduli of week sandstones. American Rock Mechanics Association. Retrieved from https://www-onepetro-org.ezproxy.lib.ucalgary.ca/conference-paper/ARMA-99-0675
Fjaer, E., Holt, R. M., Horsrud, P., Raaen, A. M., & Risnes, R. (2008). Petroleum related rock
mechanics (2nd Ed.). Amsterdam: Elsevier. Retrieved from http://app.knovel.com/hotlink/toc/id:kpPRRME001/petroleum-related-rock/petroleum-related-rock
Fossen, H., Schultz, R. A., Shipton, Z. K., & Mair, K. (2007). Deformation bands in sandstone: a
review. Journal of the Geological Society, 164, 755-769. doi: 10.1144/0016-76492006-036
141
Jaeger, J. C., Cook, N. G. W., & Zimmerman, R. W. (2007). Fundamentals of Rock Mechanics (4th ed.). Malden, MA: Blackwell Publishing.
Hirsche, K. & Hirsche, J. (2001). The use and abuse of Geostatistics. Western Geophysical.
Calgary, Canada. Retrieved from http://Library.seg.org. Isaak, E. H., & Srivastava R. M. (1989). An introduction to applied geostatistics. New York,
USA: Oxford University Press. Gonzalez, L. (2012). Numerical 3D modeling of multistage hydraulic fractures and two-way
coupling geomechanics-fluid flow simulation of a horizontal well in the Nikanassin tight gas formation. MSc Thesis, Chemical and Petroleum Engineering Department, University of Calgary, Canada.
Kantzas, A., Bryan, J., & Taheri, S. (2012). Fundamental of fluid flow in porous media.
University of Calgary, Canada. Kwon, O., Kronenberg, A. K., Gangi. A., & Johnson, B. (2001, September). Permeability of
Wilcox shale and its effective pressure law. Journal of Geophysics Research. 106(B9), 19339-19353. doi: 10.1029/2001JB000273
Li, D., Wong, L. N. Y., Liu, G., & Zhang, X. (2012). Influence of water content and anisotropy
on the strength and deformability of low porosity meta-sedimentary rocks under triaxial compression. Engineering Geology, 126, 46–66. doi: 10.1016/j.enggeo.2011.12.009.
Lopez, B., Yu, G., Aguilera, R., & Settari, A. (2014, February). Calibration of well logs with
mini-frac data for estimating the minimum horizontal stress in the tight gas Monteith Formation on the western Canada sedimentary basin: A case of Study. Society of Petroleum Engineers. doi:10.2118/168594-MS
Montmayeur, H. & Graves, R. M. (1985, January). Prediction of static elastic/mechanical
properties of consolidated and unconsolidated sands from acoustic measurements: Basic measurements. Society of Petroleum Engineers. doi:10.2118/14159-MS
Morales, R. H., & Marcinew, R. P. (1993, October). Fracturing of high-permeability formations:
Mechanical properties correlations. Society of Petroleum Engineers. doi:10.2118/26561-MS
Morita, N., Whitfill, D. L., Nygaard, O., & Bale, A. (1989, January). A Quick Method to
Determine Subsidence, Reservoir Compaction and In-Situ Stress Induced by Reservoir Depletion. Journal of Petroleum Technology. doi:10.2118/17150-PA
Mullen, M., Roundtree, R., & Barree, B. (2007, January). A composite determination of
mechanical rock properties for stimulation design (What to do when you don’t have a sonic log). Society of Petroleum Engineers. doi:10.2118/108139-MS
142
Normal Score. (n.d.). Retrieved from https://www.statsdirect.com/help/data_preparation/normal_scores.htm Ojala, I. O., & Fjær, E. (2007, January). The effective stress coefficient in porous sandstone.
American Rock Mechanics Association. Retrieved from https://www-onepetro-org.ezproxy.lib.ucalgary.ca/download/conference-paper/ARMA-07-101?id=conference-paper%2FARMA-07-101
Perkins T. K., & Weingarten J. S. (1988, January). Stability and Failure of Spherical Cavities in
Unconsolidated Sand and Weakly Consolidated Rock. Society of Petroleum Engineers. doi: 10.2118/18244-MS
Pittman, E. D. (1992). Relationship of porosity and permeability to various parameters derived
from mercury injection-capillary pressure curves for sandstone. The American Association of Petroleum Geologists. 76(2) 191-198. Retrieved from http://archives.datapages.com.ezproxy.lib.ucalgary.ca/data/bulletns/1992-93/images/pg/00760002/0000/01910.pdf
Plumb, R. A. (1994). Influence of composition and texture on the failure properties of clastic
rocks. Proceedings of Rock Mechanics in Society of Petroleum Engineers. doi: 10.2118/28022-MS.
Qiao, L. P., Wong, R. C. K., Aguilera, R., & Kantzas, A. (2009, January). Determination of
Biot’s effective-stress coefficient for permeability of Nikanassin sandstone, Petroleum Society of Canada. doi: 10.2118/2009-105
Qiu, K., Marsden, J. R., Solovyov, Y., Safdar, M., Chardac, O., & Agoco, M. S. (2005).
Downscaling geomechanics data for thin-beds using petrophysical techniques. Society of Petroleum Engineers. doi: 10.2118/93605-MS
Rider, M. & Kennedy, M. (2011). The geological interpretation of well logs (3rd Ed.). Scotland:
Rider-French Consulting Ltd. Rodriguez, H. I. (2011, January). Numerical Reservoir Simulation Coupled with Geomechanics
State of the Art and Application in Reservoir Characterization, Society of Petroleum Engineers. doi: 10.2118/152364-STU
Schlumberger Information Solutions. (2011). Advanced reservoir geomechanics with visage:
workflow/solutions training. Serra, A. L., Braz, J. A., Bechara, M. G., Palma, N. V., & Passos, C. R. (2012, January).
Reservoir geomechanics study for deepwater field identifies ways to maximize reservoir performance while reducing geomechanics risks. Society of Petroleum Engineers. doi: 10.2118/160329-MS
143
Settari, A., Walters, D., & Behie, G. (2000). Use of Coupled Reservoir and Geomechanical
Modelling for Integrated Reservoir Analysis and Management. Proceedings of Canadian International Petroleum Conference. doi: 10.2118/2000-078.
Settari, A., & Maurits, F. M. (1998, September). A Coupled Reservoir and Geomechanical
Simulation System. Society of Petroleum Engineers. doi: 10.2118/50939-PA Settari, A. T., & Walters, D. W. (1999). Reservoir Geomechanics: New approach to reservoir
engineering analysis and a powerful tool for reservoir management Field Examples, 206–208. Petroleum Society of Canada. Retrieved from http://dx.doi.org.ezproxy.lib.ucalgary.ca/10.2118/99-116
Settari, A. T., & Sen, V. (2008, January). Geomechanics in Integrated Reservoir Modeling.
Offshore Technology Conference, doi: 10.4043/19530-MS Shchipanov, A. A., Kolboth, L., & Surguchev, L. M. (2010, January). A new approach to
deformable fractured reservoir characterization: Case study of the Ekofish field. Society of Petroleum Engineers. doi: 10.2118/130425-MS
Schwall, G. H., & Denney C. A. (1994, January). Subsidence induced casing deformation
mechanisms in the Ekofisk field. Society of Petroleum Engineers. doi: 10.2118/28091-MS
Siggins, A. F., Dewhurst, D. N., Dodds, K., (2004, January). Effective stress and the Biot-Willis
coefficient for reservoir sandstones. Society of Exploration Geophysicists. Retrieved from https://www-onepetro-org.ezproxy.lib.ucalgary.ca/conference-paper/SEG-2004-1770
Soong, T. T. (2004). Fundamentals of probability and statistics for engineers. New York,
Buffalo: John Wiley & Sons. Spencer, J., Jr., Cates, M., and Thompson, D. (1994). Frame moduli of unconsolidated sands and
sandstones. Geophysics, 59(9), 1352–1361. doi: 10.1190/1.1443694. Terzaghi, K., & Beck R. B. (1949). Soil mechanics in engineering practice. John Wiley and
Sons. New York. Tran, D., Settari, A., & Nghiem, L. (2002, January). New iterative coupling between a reservoir
simulator and a geomechanics module. Society of Petroleum Engineers. doi: 10.2118/78192-MS
Trudeng, T., Garcia-Teijeiro, X., Rodriguez-Herrera, A., & Khazanehdari, J. (2014, January).
Using stochastic seismic inversion as input for 3D geomechanical models. International Petroleum Technology Conference. doi: 10.2523/17547-MS.
144
Tziallas, G. P., Saroglou, H., & Tsiambaos, G. (2013). Determination of mechanical properties
of flysch using laboratory methods. Engineering Geology, 166, 81–89. doi: 10.1016/j.enggeo.2013.09.002.
Veeken, C. A. M., Walters, J. W., Kenter, C. J., & Davies, D. R. (1989, January). Use of
plasticity models for predicting borehole stability. International Society for Rock Mechanics. Retrieved from https://www-onepetro-org.ezproxy.lib.ucalgary.ca/download/conference-paper/ISRM-IS-1989-106?id=conference-paper%2FISRM-IS-1989-106.
Vernik, L., Bruno, M., & Bovberg, C. (1993). Empirical relations between compressive strength
and porosity of siliciclastic rocks. International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 30(7), 677–680. doi: 10.1016/0148-9062(93)90004-W.
Vincke, O., Longuemare, P., Bouteca, M., & Deflandre, J. P. (1998). Investigation of the
poromechanical behaviour of shales in the elastic domain. Proceedings of SPE/ISRM Rock Mechanics in Petroleum Engineering. doi: 10.2118/47589-MS.
Wang, Z. N. (2000). Seismic and acoustic velocities in reservoir rocks: Recent developments.
vol. 3: Tulsa: Society of Exploration Geophysicists. White, J. E. (1991). Underground sound: Appled seismic waves: Amstedam: Elseiver. Wood, D., & Shaw, N. (2012, May). The correlation of unconfined compressive strength with
dry density and grainsize for different aged sedimentary rock. International Society for Rock Mechanics. Retrieved from https://www-onepetro-org.ezproxy.lib.ucalgary.ca/conference-paper/ISRM-EUROCK-2012-111
Wu, R. S. (1994). Wide-angle elastic waves one-way propagation in heterogeneous media an
elastic wave complex-screen method. Journal of Geophysical Research, 99(B1), 751-766. Retrieved from http://webapps2.ucalgary.ca/~lcrtest/search/index.php?q=Wide-angle+elastic+waves+one-way+propagation
Yale, D. P., & Jamieson, W. H. (1994). Static and dynamic mechanical properties of carbonates.
American Rock Mechanics Association. Retrieved from https://www-onepetro-org.ezproxy.lib.ucalgary.ca/conference-paper/ARMA-1994-0463
Zhang, Y. G., Wang, X. J., Zhang, Z. X., Fan, K. M., & Yu, Y. M. (2010, January). Predicting
inter-well rock mechanics parameters using geophysical logs and 3-D seismic data. American Rock Mechanics Association. Retrieved from https://www-onepetro-org.ezproxy.lib.ucalgary.ca/conference-paper/ARMA-07-016
145
Zoback, M. D. (2006). Reservoir Geomechanics: Earth stress and rock mechanics applied to exploration, production and wellbore stability. Standford: Cambridge University Press.
Appendix A: Geostatistical Data Analysis
A.1 Data Analysis Models
Geostatistical characterization techniques help to relate petrophysical properties
measured at the wells and describe their spatial continuity. These techniques provide estimations
of properties where sample data are not available and help predicting uncertainties associated
with those estimations.
Isaaks and Srivastava (1989) indicated that geostatistical data analysis is based on the fact
that a semivariogram can describe the spatial structure of a variable. A semivariogram is defined
as half the average square distance between a pair of data values separated by a distance h. This
is expressed by Eq. A-1 .
𝛾𝛾(ℎ) =1
2𝑁𝑁(ℎ) �[𝑧𝑧(𝑥𝑥𝑚𝑚 − ℎ) − 𝑧𝑧(𝑥𝑥𝑥𝑥)]2
𝑁𝑁(ℎ)
𝑚𝑚=1
Eq. A-1
Where N(h) represents the number of pairs separated by a distance h, also called lag.
Isaaks and Srivastava (1989) also presented some useful concepts related to
semivariograms that can be describes as follows:
Nugget: It is used to introduce a discontinuity at the origin of a semivariogram model.
Explanations for this discontinuity comprise sampling short scale variability and error.
Sill: Maximum semivariance value observed in the spatial structure for a variable.
Range: The distance at which the data are no longer correlated.
The necessity of a semivariogram model resides in the fact that a semivariogram value
for a specific distance or orientation cannot be available in the sample semivariogram values.
146
There are two types of variograms, those that never reach a sill and those that do. The models
that reach a sill are basically spherical, exponential and/or Gaussian models. Occasionally it is
necessary to combine these models to have a representation for the experimental semivariogram
sample.
The spherical model is very common and is expressed by Eq. A-2:
𝛾𝛾(ℎ) = 1.5ℎ𝑎𝑎− 0.5 �
ℎ𝑎𝑎�3
Eq. A-2
The exponential model is another common method. It is represented by Eq. A-3:
𝛾𝛾(ℎ) = 1 − exp �−3ℎ𝑎𝑎� Eq. A-3
The Gaussian model is used particularly to handle continuous phenomena. Its
standardized expression is given by Eq. A-4:
𝛾𝛾(ℎ) = 1 − exp �−3ℎ2
𝑎𝑎2� Eq. A-4
Where a is the range
A.2 Distribution of Properties
The estimation of petrophysical properties in a 3D MEM model is developed by applying
the following three steps: (1) Scale-up the well logs in the cell crossed by the well trajectory. (2)
Determine semivariogram models in the vertical direction and in the major and minor directions.
(3) Estimate petrophysical properties in the models using the semivariograms and the stochastic
Sequential Gaussian Simulation. This process was applied for estimating properties in the
sideburden and overburden. The following figures summarize the results by property and zone.
147
A.2.1 Young’s modulus
Zone 2 (Upper Pliocene)
Figure A-1 Black squares represent the experimental data, the black line the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction,
middle variogram in major direction and lower variogram in minor direction).
149
Figure A-2 3D MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between results from well logs used for
upscaling, the upscaled well log in the cell crossed by the well and the cells in the model.
Mean=2.07 GPaStd. dev. =0.79 Gpa
150
Zone 3 (MI-2)
Figure A-3 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
151
Figure A-4 3D MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing comparison between the well log used for upscaling,
the upscaled well log in the cell crossed by the well and the cells in the model.
Mean=2.07 GPaStd. dev. =0.79 Gpa
152
Zone 5 (MI-1)
Figure A-5 Black squares represent the experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction,
middle variogram in major direction and lower variogram in minor direction).
153
Figure A-6 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=2.45 GPaStd. dev. =0.65 Gpa
154
A.2.2 Poisson’s ratio
Zone 2 (Upper Pliocene)
Figure A-7 Black squares represent the experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction,
middle variogram in major direction and lower variogram in minor direction).
155
Figure A-8 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model..
Mean=0.25Std. dev. =0.018
156
Zone 3 (MI-2)
Figure A-9 Black squares represent the experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction,
middle variogram in major direction and lower variogram in minor direction).
157
Figure A-10 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=0.23Std. dev. =0.018
158
Zone 5 (MI-1)
Figure A-11 Black squares represent the experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction,
middle variogram in major direction and lower variogram in minor direction).
159
Figure A-12 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=0.22Std. dev. =0.018
160
A.2.3 UCS
Zone 2 (Upper Pliocene)
Figure A-13 Black squares represent the experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction,
middle variogram in major direction and lower variogram in minor direction).
161
Figure A-14 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=4.77 MPaStd. dev. =1.67 MPa
162
Zone 3 (MI-2)
Figure A-15 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
163
Figure A-16 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=8.03 MPaStd. dev. =2.40 MPa
164
Zone 5 (MI-1)
Figure A-17 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
165
Figure A-18 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=9.52 MPaStd. dev. =2.0 MPa
166
A.2.4 Porosity
Zone 2 (Upper Pliocene)
Figure A-19 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
167
Figure A-20 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=0.126 m3/m3
Std. dev. =0.08 m3/m3
168
Zone 3 (MI-2)
Figure A-21 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
169
Figure A-22 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=0.14 m3/m3
Std. dev. =0.06 m3/m3
170
Zone 5 (MI-1)
Figure A-23 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
171
Figure A-24 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=0.12 m3/m3
Std. dev. =0.06 m3/m3
172
A.2.5 Bulk density
Zone 2 (Upper Pliocene)
Figure A-25 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
173
Figure A-26 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=2.19 g/m3
Std. dev. =0.11 g/m3
174
Zone 3 (MI-2)
Figure A-27 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
175
Figure A-28 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=2.20 g/m3
Std. dev. =0.08 g/m3
176
Zone 5 (MI-1)
Figure A-29 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
177
Figure A-30 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=2.27 g/m3
Std. dev. =0.07 g/m3
178
A.2.6 Friction angle
Zone 2 (Upper Pliocene)
Figure A-31 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
179
Figure A-32 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=31.9 degreesStd. dev.=6.26 degrees
180
Zone 3 (MI-2)
Figure A-33 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
181
Figure A-34 MEM property distribution. (Top graph) Property distribution in the model. (Bottom) Histogram showing a comparison between the well log used for upscaling, the
upscaled well log in the cell crossed by the well and the cells in the model.
Mean=34.7 degreesStd. dev.=4.89 degrees
182
Zone 5 (MI-1)
Figure A-35 Black squares represent experimental data, the black line is the regression variogram and the blue line the variogram model. (Upper variogram in vertical direction
middle variogram in major direction and lower variogram in minor direction).
183
Top Related